Datasets:

elplaguister

/

DTS_pair_datasets

like 0

Oh , oh. I see.

Mm - hmm.

if it gets an unambiguous result then you 're definitely in a in a in a voice in a , uh , s region with speech. Uh.

So there 's this assumption that the v the voice activity detector can only use the MFCC ?

QMSum_120

Mm - hmm.

if it gets an unambiguous result then you 're definitely in a in a in a voice in a , uh , s region with speech. Uh.

So there 's this assumption that the v the voice activity detector can only use the MFCC ?

That 's not clear , but this e

QMSum_120

if it gets an unambiguous result then you 're definitely in a in a in a voice in a , uh , s region with speech. Uh.

So there 's this assumption that the v the voice activity detector can only use the MFCC ?

That 's not clear , but this e

Well , for the baseline.

QMSum_120

So there 's this assumption that the v the voice activity detector can only use the MFCC ?

That 's not clear , but this e

Well , for the baseline.

Yeah.

QMSum_120

That 's not clear , but this e

Well , for the baseline.

Yeah.

So so if you use other features then y But it 's just a question of what is your baseline. Right ? What is it that you 're supposed to do better than ?

QMSum_120

Well , for the baseline.

Yeah.

So so if you use other features then y But it 's just a question of what is your baseline. Right ? What is it that you 're supposed to do better than ?

I g Yeah.

QMSum_120

Yeah.

So so if you use other features then y But it 's just a question of what is your baseline. Right ? What is it that you 're supposed to do better than ?

I g Yeah.

And so having the baseline be the MFCC 's means that people could choose to pour their ener their effort into trying to do a really good VAD

QMSum_120

So so if you use other features then y But it 's just a question of what is your baseline. Right ? What is it that you 're supposed to do better than ?

I g Yeah.

And so having the baseline be the MFCC 's means that people could choose to pour their ener their effort into trying to do a really good VAD

I don't s But they seem like two separate issues.

QMSum_120

I g Yeah.

And so having the baseline be the MFCC 's means that people could choose to pour their ener their effort into trying to do a really good VAD

I don't s But they seem like two separate issues.

or tryi They 're sort of separate.

QMSum_120

And so having the baseline be the MFCC 's means that people could choose to pour their ener their effort into trying to do a really good VAD

I don't s But they seem like two separate issues.

or tryi They 're sort of separate.

Right ? I mean

QMSum_120

I don't s But they seem like two separate issues.

or tryi They 're sort of separate.

Right ? I mean

Unfortunately there 's coupling between them , which is part of what I think Stephane is getting to , is that you can choose your features in such a way as to improve the VAD.

QMSum_120

or tryi They 're sort of separate.

Right ? I mean

Unfortunately there 's coupling between them , which is part of what I think Stephane is getting to , is that you can choose your features in such a way as to improve the VAD.

Yeah.

QMSum_120

Right ? I mean

Unfortunately there 's coupling between them , which is part of what I think Stephane is getting to , is that you can choose your features in such a way as to improve the VAD.

Yeah.

And you also can choose your features in such a way as to prove improve recognition. They may not be the same thing.

QMSum_120

Unfortunately there 's coupling between them , which is part of what I think Stephane is getting to , is that you can choose your features in such a way as to improve the VAD.

Yeah.

And you also can choose your features in such a way as to prove improve recognition. They may not be the same thing.

But it seems like you should do both.

QMSum_120

Yeah.

And you also can choose your features in such a way as to prove improve recognition. They may not be the same thing.

But it seems like you should do both.

You should do both

QMSum_120

And you also can choose your features in such a way as to prove improve recognition. They may not be the same thing.

But it seems like you should do both.

You should do both

Right ?

QMSum_120

But it seems like you should do both.

You should do both

Right ?

and and I I think that this still makes I still think this makes sense as a baseline. It 's just saying , as a baseline , we know

QMSum_120

You should do both

Right ?

and and I I think that this still makes I still think this makes sense as a baseline. It 's just saying , as a baseline , we know

Mmm.

QMSum_120

Right ?

and and I I think that this still makes I still think this makes sense as a baseline. It 's just saying , as a baseline , we know

Mmm.

you know , we had the MFCC 's before , lots of people have done voice activity detectors ,

QMSum_120

and and I I think that this still makes I still think this makes sense as a baseline. It 's just saying , as a baseline , we know

Mmm.

you know , we had the MFCC 's before , lots of people have done voice activity detectors ,

Mm - hmm.

QMSum_120

Mmm.

you know , we had the MFCC 's before , lots of people have done voice activity detectors ,

Mm - hmm.

you might as well pick some voice activity detector and make that the baseline , just like you picked some version of HTK and made that the baseline.

QMSum_120

you know , we had the MFCC 's before , lots of people have done voice activity detectors ,

Mm - hmm.

you might as well pick some voice activity detector and make that the baseline , just like you picked some version of HTK and made that the baseline.

Yeah. Right.

QMSum_120

Mm - hmm.

you might as well pick some voice activity detector and make that the baseline , just like you picked some version of HTK and made that the baseline.

Yeah. Right.

And then let 's try and make everything better. Um , and if one of the ways you make it better is by having your features be better features for the VAD then that 's so be it.

QMSum_120

you might as well pick some voice activity detector and make that the baseline , just like you picked some version of HTK and made that the baseline.

Yeah. Right.

And then let 's try and make everything better. Um , and if one of the ways you make it better is by having your features be better features for the VAD then that 's so be it.

Mm - hmm.

QMSum_120

Yeah. Right.

And then let 's try and make everything better. Um , and if one of the ways you make it better is by having your features be better features for the VAD then that 's so be it.

Mm - hmm.

But , uh , uh , uh , at least you have a starting point that 's um , cuz i i some of the some of the people didn't have a VAD at all , I guess. Right ? And and

QMSum_120

And then let 's try and make everything better. Um , and if one of the ways you make it better is by having your features be better features for the VAD then that 's so be it.

Mm - hmm.

But , uh , uh , uh , at least you have a starting point that 's um , cuz i i some of the some of the people didn't have a VAD at all , I guess. Right ? And and

Yeah.

QMSum_120

Mm - hmm.

But , uh , uh , uh , at least you have a starting point that 's um , cuz i i some of the some of the people didn't have a VAD at all , I guess. Right ? And and

Yeah.

then they they looked pretty bad and and in fact what they were doing wasn't so bad at all.

QMSum_120

But , uh , uh , uh , at least you have a starting point that 's um , cuz i i some of the some of the people didn't have a VAD at all , I guess. Right ? And and

Yeah.

then they they looked pretty bad and and in fact what they were doing wasn't so bad at all.

Mm - hmm. Mm - hmm.

QMSum_120

Yeah.

then they they looked pretty bad and and in fact what they were doing wasn't so bad at all.

Mm - hmm. Mm - hmm.

But , um.

QMSum_120

then they they looked pretty bad and and in fact what they were doing wasn't so bad at all.

Mm - hmm. Mm - hmm.

But , um.

Yeah. It seems like you should try to make your baseline as good as possible. And if it turns out that you can't improve on that , well , I mean , then , you know , nobody wins and you just use MFCC. Right ?

QMSum_120

Mm - hmm. Mm - hmm.

But , um.

Yeah. It seems like you should try to make your baseline as good as possible. And if it turns out that you can't improve on that , well , I mean , then , you know , nobody wins and you just use MFCC. Right ?

Yeah. I mean , it seems like , uh , it should include sort of the current state of the art that you want are trying to improve , and MFCC 's , you know , or PLP or something it seems like reasonable baseline for the features , and anybody doing this task , uh , is gonna have some sort of voice activity detection at some level , in some way. They might use the whole recognizer to do it but rather than a separate thing , but but they 'll have it on some level. So , um.

QMSum_120

But , um.

Yeah. It seems like you should try to make your baseline as good as possible. And if it turns out that you can't improve on that , well , I mean , then , you know , nobody wins and you just use MFCC. Right ?

Yeah. I mean , it seems like , uh , it should include sort of the current state of the art that you want are trying to improve , and MFCC 's , you know , or PLP or something it seems like reasonable baseline for the features , and anybody doing this task , uh , is gonna have some sort of voice activity detection at some level , in some way. They might use the whole recognizer to do it but rather than a separate thing , but but they 'll have it on some level. So , um.

It seems like whatever they choose they shouldn't , you know , purposefully brain - damage a part of the system to make a worse baseline , or

QMSum_120

Yeah. It seems like you should try to make your baseline as good as possible. And if it turns out that you can't improve on that , well , I mean , then , you know , nobody wins and you just use MFCC. Right ?

Yeah. I mean , it seems like , uh , it should include sort of the current state of the art that you want are trying to improve , and MFCC 's , you know , or PLP or something it seems like reasonable baseline for the features , and anybody doing this task , uh , is gonna have some sort of voice activity detection at some level , in some way. They might use the whole recognizer to do it but rather than a separate thing , but but they 'll have it on some level. So , um.

It seems like whatever they choose they shouldn't , you know , purposefully brain - damage a part of the system to make a worse baseline , or

Well , I think people just had

QMSum_120

Yeah. I mean , it seems like , uh , it should include sort of the current state of the art that you want are trying to improve , and MFCC 's , you know , or PLP or something it seems like reasonable baseline for the features , and anybody doing this task , uh , is gonna have some sort of voice activity detection at some level , in some way. They might use the whole recognizer to do it but rather than a separate thing , but but they 'll have it on some level. So , um.

It seems like whatever they choose they shouldn't , you know , purposefully brain - damage a part of the system to make a worse baseline , or

Well , I think people just had

You know ?

QMSum_120

It seems like whatever they choose they shouldn't , you know , purposefully brain - damage a part of the system to make a worse baseline , or

Well , I think people just had

You know ?

it wasn't that they purposely brain - damaged it. I think people hadn't really thought through about the , uh the VAD issue.

QMSum_120

Well , I think people just had

You know ?

it wasn't that they purposely brain - damaged it. I think people hadn't really thought through about the , uh the VAD issue.

Mmm.

QMSum_120

You know ?

it wasn't that they purposely brain - damaged it. I think people hadn't really thought through about the , uh the VAD issue.

Mmm.

Mm - hmm.

QMSum_120

it wasn't that they purposely brain - damaged it. I think people hadn't really thought through about the , uh the VAD issue.

Mmm.

Mm - hmm.

And and then when the the the proposals actually came in and half of them had V A Ds and half of them didn't , and the half that did did well and the half that didn't did poorly.

QMSum_120

Mmm.

Mm - hmm.

And and then when the the the proposals actually came in and half of them had V A Ds and half of them didn't , and the half that did did well and the half that didn't did poorly.

Mm - hmm.

QMSum_120

Mm - hmm.

And and then when the the the proposals actually came in and half of them had V A Ds and half of them didn't , and the half that did did well and the half that didn't did poorly.

Mm - hmm.

So it 's

QMSum_120

And and then when the the the proposals actually came in and half of them had V A Ds and half of them didn't , and the half that did did well and the half that didn't did poorly.

Mm - hmm.

So it 's

Mm - hmm. Um.

QMSum_120

Mm - hmm.

So it 's

Mm - hmm. Um.

Uh.

QMSum_120

So it 's

Mm - hmm. Um.

Uh.

Yeah. So we 'll see what happen with this. And Yeah. So what happened since , um , last week is well , from OGI , these experiments on putting VAD on the baseline. And these experiments also are using , uh , some kind of noise compensation , so spectral subtraction , and putting on - line normalization , um , just after this. So I think spectral subtraction , LDA filtering , and on - line normalization , so which is similar to the pro proposal - one , but with spectral subtraction in addition , and it seems that on - line normalization doesn't help further when you have spectral subtraction.

QMSum_120

Mm - hmm. Um.

Uh.

Yeah. So we 'll see what happen with this. And Yeah. So what happened since , um , last week is well , from OGI , these experiments on putting VAD on the baseline. And these experiments also are using , uh , some kind of noise compensation , so spectral subtraction , and putting on - line normalization , um , just after this. So I think spectral subtraction , LDA filtering , and on - line normalization , so which is similar to the pro proposal - one , but with spectral subtraction in addition , and it seems that on - line normalization doesn't help further when you have spectral subtraction.

Is this related to the issue that you brought up a couple of meetings ago with the the musical tones

QMSum_120

Uh.

Yeah. So we 'll see what happen with this. And Yeah. So what happened since , um , last week is well , from OGI , these experiments on putting VAD on the baseline. And these experiments also are using , uh , some kind of noise compensation , so spectral subtraction , and putting on - line normalization , um , just after this. So I think spectral subtraction , LDA filtering , and on - line normalization , so which is similar to the pro proposal - one , but with spectral subtraction in addition , and it seems that on - line normalization doesn't help further when you have spectral subtraction.

Is this related to the issue that you brought up a couple of meetings ago with the the musical tones

and ?

QMSum_120

Yeah. So we 'll see what happen with this. And Yeah. So what happened since , um , last week is well , from OGI , these experiments on putting VAD on the baseline. And these experiments also are using , uh , some kind of noise compensation , so spectral subtraction , and putting on - line normalization , um , just after this. So I think spectral subtraction , LDA filtering , and on - line normalization , so which is similar to the pro proposal - one , but with spectral subtraction in addition , and it seems that on - line normalization doesn't help further when you have spectral subtraction.

Is this related to the issue that you brought up a couple of meetings ago with the the musical tones

and ?

I have no idea , because the issue I brought up was with a very simple spectral subtraction approach ,

QMSum_120

Is this related to the issue that you brought up a couple of meetings ago with the the musical tones

and ?

I have no idea , because the issue I brought up was with a very simple spectral subtraction approach ,

Mmm.

QMSum_120

and ?

I have no idea , because the issue I brought up was with a very simple spectral subtraction approach ,

Mmm.

and the one that they use at OGI is one from from the proposed the the the Aurora prop uh , proposals , which might be much better. So , yeah. I asked Sunil for more information about that , but , uh , I don't know yet. Um. And what 's happened here is that we so we have this kind of new , um , reference system which use a nice a a clean downsampling - upsampling , which use a new filter that 's much shorter and which also cuts the frequency below sixty - four hertz ,

QMSum_120

I have no idea , because the issue I brought up was with a very simple spectral subtraction approach ,

Mmm.

and the one that they use at OGI is one from from the proposed the the the Aurora prop uh , proposals , which might be much better. So , yeah. I asked Sunil for more information about that , but , uh , I don't know yet. Um. And what 's happened here is that we so we have this kind of new , um , reference system which use a nice a a clean downsampling - upsampling , which use a new filter that 's much shorter and which also cuts the frequency below sixty - four hertz ,

Right.

QMSum_120

Mmm.

and the one that they use at OGI is one from from the proposed the the the Aurora prop uh , proposals , which might be much better. So , yeah. I asked Sunil for more information about that , but , uh , I don't know yet. Um. And what 's happened here is that we so we have this kind of new , um , reference system which use a nice a a clean downsampling - upsampling , which use a new filter that 's much shorter and which also cuts the frequency below sixty - four hertz ,

Right.

which was not done on our first proposal.

QMSum_120

and the one that they use at OGI is one from from the proposed the the the Aurora prop uh , proposals , which might be much better. So , yeah. I asked Sunil for more information about that , but , uh , I don't know yet. Um. And what 's happened here is that we so we have this kind of new , um , reference system which use a nice a a clean downsampling - upsampling , which use a new filter that 's much shorter and which also cuts the frequency below sixty - four hertz ,

Right.

which was not done on our first proposal.

When you say " we have that " , does Sunil have it now , too ,

QMSum_120

Right.

which was not done on our first proposal.

When you say " we have that " , does Sunil have it now , too ,

I No.

QMSum_120

which was not done on our first proposal.

When you say " we have that " , does Sunil have it now , too ,

I No.

or ?

QMSum_120

When you say " we have that " , does Sunil have it now , too ,

I No.

or ?

No.

QMSum_120

I No.

or ?

No.

OK.

QMSum_120

or ?

No.

OK.

Because we 're still testing. So we have the result for , uh , just the features

QMSum_120

No.

OK.

Because we 're still testing. So we have the result for , uh , just the features

OK.

QMSum_120

OK.

Because we 're still testing. So we have the result for , uh , just the features

OK.

and we are currently testing with putting the neural network in the KLT. Um , it seems to improve on the well - matched case , um , but it 's a little bit worse on the mismatch and highly - mismatched I mean when we put the neural network. And with the current weighting I think it 's sh it will be better because the well - matched case is better. Mmm.

QMSum_120

Because we 're still testing. So we have the result for , uh , just the features

OK.

and we are currently testing with putting the neural network in the KLT. Um , it seems to improve on the well - matched case , um , but it 's a little bit worse on the mismatch and highly - mismatched I mean when we put the neural network. And with the current weighting I think it 's sh it will be better because the well - matched case is better. Mmm.

But how much worse since the weighting might change how how much worse is it on the other conditions , when you say it 's a little worse ?

QMSum_120

OK.

and we are currently testing with putting the neural network in the KLT. Um , it seems to improve on the well - matched case , um , but it 's a little bit worse on the mismatch and highly - mismatched I mean when we put the neural network. And with the current weighting I think it 's sh it will be better because the well - matched case is better. Mmm.

But how much worse since the weighting might change how how much worse is it on the other conditions , when you say it 's a little worse ?

It 's like , uh , fff , fff um , ten percent relative. Yeah.

QMSum_120

and we are currently testing with putting the neural network in the KLT. Um , it seems to improve on the well - matched case , um , but it 's a little bit worse on the mismatch and highly - mismatched I mean when we put the neural network. And with the current weighting I think it 's sh it will be better because the well - matched case is better. Mmm.

But how much worse since the weighting might change how how much worse is it on the other conditions , when you say it 's a little worse ?

It 's like , uh , fff , fff um , ten percent relative. Yeah.

OK. Um.

QMSum_120

But how much worse since the weighting might change how how much worse is it on the other conditions , when you say it 's a little worse ?

It 's like , uh , fff , fff um , ten percent relative. Yeah.

OK. Um.

Mm - hmm.

QMSum_120

It 's like , uh , fff , fff um , ten percent relative. Yeah.

OK. Um.

Mm - hmm.

But it has the , uh the latencies are much shorter. That 's

QMSum_120

OK. Um.

Mm - hmm.

But it has the , uh the latencies are much shorter. That 's

Uh - y w when I say it 's worse , it 's not it 's when I I uh , compare proposal - two to proposal - one , so , r uh , y putting neural network compared to n not having any neural network. I mean , this new system is is is better ,

QMSum_120

Mm - hmm.

But it has the , uh the latencies are much shorter. That 's

Uh - y w when I say it 's worse , it 's not it 's when I I uh , compare proposal - two to proposal - one , so , r uh , y putting neural network compared to n not having any neural network. I mean , this new system is is is better ,

Uh - huh.

QMSum_120

But it has the , uh the latencies are much shorter. That 's

Uh - y w when I say it 's worse , it 's not it 's when I I uh , compare proposal - two to proposal - one , so , r uh , y putting neural network compared to n not having any neural network. I mean , this new system is is is better ,

Uh - huh.

because it has um , this sixty - four hertz cut - off , uh , clean downsampling , and , um what else ? Uh , yeah , a good VAD. We put the good VAD. So. Yeah , I don't know. I I j uh , uh pr

QMSum_120

Uh - y w when I say it 's worse , it 's not it 's when I I uh , compare proposal - two to proposal - one , so , r uh , y putting neural network compared to n not having any neural network. I mean , this new system is is is better ,

Uh - huh.

because it has um , this sixty - four hertz cut - off , uh , clean downsampling , and , um what else ? Uh , yeah , a good VAD. We put the good VAD. So. Yeah , I don't know. I I j uh , uh pr

But the latencies but you 've got the latency shorter now.

QMSum_120

Uh - huh.

because it has um , this sixty - four hertz cut - off , uh , clean downsampling , and , um what else ? Uh , yeah , a good VAD. We put the good VAD. So. Yeah , I don't know. I I j uh , uh pr

But the latencies but you 've got the latency shorter now.

Latency is short is Yeah.

QMSum_120

because it has um , this sixty - four hertz cut - off , uh , clean downsampling , and , um what else ? Uh , yeah , a good VAD. We put the good VAD. So. Yeah , I don't know. I I j uh , uh pr

But the latencies but you 've got the latency shorter now.

Latency is short is Yeah.

Yeah.

QMSum_120

But the latencies but you 've got the latency shorter now.

Latency is short is Yeah.

Yeah.

Isn't it

QMSum_120

Latency is short is Yeah.

Yeah.

Isn't it

And so

QMSum_120

Yeah.

Isn't it

And so

So it 's better than the system that we had before.

QMSum_120

Isn't it

And so

So it 's better than the system that we had before.

Yeah. Mainly because of the sixty - four hertz and the good VAD.

QMSum_120

And so

So it 's better than the system that we had before.

Yeah. Mainly because of the sixty - four hertz and the good VAD.

OK.

QMSum_120

So it 's better than the system that we had before.

Yeah. Mainly because of the sixty - four hertz and the good VAD.

OK.

And then I took this system and , mmm , w uh , I p we put the old filters also. So we have this good system , with good VAD , with the short filter and with the long filter , and , um , with the short filter it 's not worse. So well , is it

QMSum_120

Yeah. Mainly because of the sixty - four hertz and the good VAD.

OK.

And then I took this system and , mmm , w uh , I p we put the old filters also. So we have this good system , with good VAD , with the short filter and with the long filter , and , um , with the short filter it 's not worse. So well , is it

OK.

QMSum_120

OK.

And then I took this system and , mmm , w uh , I p we put the old filters also. So we have this good system , with good VAD , with the short filter and with the long filter , and , um , with the short filter it 's not worse. So well , is it

OK.

it 's in

QMSum_120

And then I took this system and , mmm , w uh , I p we put the old filters also. So we have this good system , with good VAD , with the short filter and with the long filter , and , um , with the short filter it 's not worse. So well , is it

OK.

it 's in

So that 's that 's all fine.

QMSum_120

OK.

it 's in

So that 's that 's all fine.

Yes. Uh

QMSum_120

it 's in

So that 's that 's all fine.

Yes. Uh

But what you 're saying is that when you do these So let me try to understand. When when you do these same improvements to proposal - one ,

QMSum_120

So that 's that 's all fine.

Yes. Uh

But what you 're saying is that when you do these So let me try to understand. When when you do these same improvements to proposal - one ,

Mm - hmm.

QMSum_120

Yes. Uh

But what you 're saying is that when you do these So let me try to understand. When when you do these same improvements to proposal - one ,

Mm - hmm.

that , uh , on the i things are somewhat better , uh , in proposal - two for the well - matched case and somewhat worse for the other two cases.

QMSum_120

But what you 're saying is that when you do these So let me try to understand. When when you do these same improvements to proposal - one ,

Mm - hmm.

that , uh , on the i things are somewhat better , uh , in proposal - two for the well - matched case and somewhat worse for the other two cases.

Yeah.

QMSum_120

Mm - hmm.

that , uh , on the i things are somewhat better , uh , in proposal - two for the well - matched case and somewhat worse for the other two cases.

Yeah.

So does , uh when you say , uh So The th now that these other things are in there , is it the case maybe that the additions of proposal - two over proposal - one are less im important ?

QMSum_120

that , uh , on the i things are somewhat better , uh , in proposal - two for the well - matched case and somewhat worse for the other two cases.

Yeah.

So does , uh when you say , uh So The th now that these other things are in there , is it the case maybe that the additions of proposal - two over proposal - one are less im important ?

Yeah. Probably , yeah.

QMSum_120

Yeah.

So does , uh when you say , uh So The th now that these other things are in there , is it the case maybe that the additions of proposal - two over proposal - one are less im important ?

Yeah. Probably , yeah.

I get it.

QMSum_120

So does , uh when you say , uh So The th now that these other things are in there , is it the case maybe that the additions of proposal - two over proposal - one are less im important ?

Yeah. Probably , yeah.

I get it.

Um So , yeah. Uh. Yeah , but it 's a good thing anyway to have shorter delay. Then we tried , um , to do something like proposal - two but having , um , e using also MSG features. So there is this KLT part , which use just the standard features ,

QMSum_120

Yeah. Probably , yeah.

I get it.

Um So , yeah. Uh. Yeah , but it 's a good thing anyway to have shorter delay. Then we tried , um , to do something like proposal - two but having , um , e using also MSG features. So there is this KLT part , which use just the standard features ,

Mm - hmm. Right.

QMSum_120

I get it.

Um So , yeah. Uh. Yeah , but it 's a good thing anyway to have shorter delay. Then we tried , um , to do something like proposal - two but having , um , e using also MSG features. So there is this KLT part , which use just the standard features ,

Mm - hmm. Right.

and then two neura two neural networks.

QMSum_120

Um So , yeah. Uh. Yeah , but it 's a good thing anyway to have shorter delay. Then we tried , um , to do something like proposal - two but having , um , e using also MSG features. So there is this KLT part , which use just the standard features ,

Mm - hmm. Right.

and then two neura two neural networks.

Mm - hmm.

QMSum_120

Mm - hmm. Right.

and then two neura two neural networks.

Mm - hmm.

Mmm , and it doesn't seem to help. Um , however , we just have one result , which is the Italian mismatch , so. Uh. We have to wait for that to fill the whole table , but

QMSum_120

and then two neura two neural networks.

Mm - hmm.

Mmm , and it doesn't seem to help. Um , however , we just have one result , which is the Italian mismatch , so. Uh. We have to wait for that to fill the whole table , but

OK. There was a start of some effort on something related to voicing or something. Is that ?

QMSum_120

Mm - hmm.

Mmm , and it doesn't seem to help. Um , however , we just have one result , which is the Italian mismatch , so. Uh. We have to wait for that to fill the whole table , but

OK. There was a start of some effort on something related to voicing or something. Is that ?

Yeah. Um , yeah. So basically we try to , uh , find good features that could be used for voicing detection , uh , but it 's still , uh on the , um t

QMSum_120

Mmm , and it doesn't seem to help. Um , however , we just have one result , which is the Italian mismatch , so. Uh. We have to wait for that to fill the whole table , but

OK. There was a start of some effort on something related to voicing or something. Is that ?

Yeah. Um , yeah. So basically we try to , uh , find good features that could be used for voicing detection , uh , but it 's still , uh on the , um t

Oh , well , I have the picture.

QMSum_120

OK. There was a start of some effort on something related to voicing or something. Is that ?

Yeah. Um , yeah. So basically we try to , uh , find good features that could be used for voicing detection , uh , but it 's still , uh on the , um t

Oh , well , I have the picture.

we w basically we are still playing with Matlab to to look at at what happened ,

QMSum_120

Yeah. Um , yeah. So basically we try to , uh , find good features that could be used for voicing detection , uh , but it 's still , uh on the , um t

Oh , well , I have the picture.

we w basically we are still playing with Matlab to to look at at what happened ,

What sorts of

QMSum_120

Oh , well , I have the picture.

we w basically we are still playing with Matlab to to look at at what happened ,

What sorts of

Yeah.

QMSum_120

we w basically we are still playing with Matlab to to look at at what happened ,

What sorts of

Yeah.

and

QMSum_120

What sorts of

Yeah.

and

what sorts of features are you looking at ?

QMSum_120

Yeah.

and

what sorts of features are you looking at ?

We have some

QMSum_120