text stringlengths 61 260 |
|---|
10:22:59 [offline-stt] 3.84s audio -> 341ms decode two-way -> vi (vi syllables 100% of 2, syllables) vi="Ô KÊ" en="Okay, title." |
10:22:59 [offline-bridge] #0 (endpoint) vi->en: "Ô kê" => "Oh my" |
10:23:02 [offline-stt] 2.36s audio -> 215ms decode two-way -> vi (vi syllables 100% of 4, syllables) vi="NÀY NHƯ NÀO NHỜ" en="It's clean now, yeah." |
10:23:02 [offline-bridge] #1 (endpoint) vi->en: "Này như nào nhờ" => "How is it today?" |
10:23:04 [offline-stt] 1.46s audio -> 181ms decode two-way -> en (vi syllables 0% of 0, syllables) vi="" en="They don't move." |
10:23:05 [offline-bridge] #2 (endpoint) en->vi: "They don't move." => "Họ không di chuyển." |
10:23:15 [offline-stt] 1.06s audio -> 130ms decode two-way -> en (vi syllables 0% of 0, syllables) vi="" en="Mm-hmm." |
10:23:15 [offline-bridge] #3 (endpoint) en->vi: "Mm-hmm." => "Ừm." |
10:23:16 [offline-stt] 1.48s audio -> 165ms decode two-way -> vi (vi syllables 100% of 3, syllables) vi="NHÌN ĐÂY LÀ" en="Yeah." |
10:23:17 [offline-bridge] #4 (endpoint) vi->en: "Nhìn đây là" => "Look here is" |
10:23:20 [offline-stt] 1.76s audio -> 173ms decode two-way -> vi (vi syllables 100% of 2, syllables) vi="Ờ THÌ" en="Thirteen" |
10:23:20 [offline-bridge] #5 (endpoint) vi->en: "Ờ thì" => "Well then" |
10:23:24 [offline-stt] 2.50s audio -> 207ms decode two-way -> vi (vi syllables 75% of 4, syllables) vi="ĐỂ CHỖ COPI SANG" en="Namari to opinions and" |
10:23:24 [offline-bridge] #6 (endpoint) vi->en: "Để chỗ copi sang" => "For copying to the destination" |
10:23:31 [offline-stt] 2.92s audio -> 229ms decode two-way -> vi (vi syllables 89% of 9, syllables) vi="SAO ANH KHÔNG VÔ CÁI OVERLIP CUP BI LUÔN" en="Shan hom yoke ovalik kabai do" |
10:23:32 [offline-bridge] #7 (endpoint) vi->en: "Sao anh không vô cái overlip cup bi luôn" => "Why don't you just go to that overlip cup?" |
10:23:35 [offline-stt] 3.80s audio -> 308ms decode two-way -> en (vi syllables 71% of 7, syllables) vi="Ý NÀY KHÔNG PHẢI LÀ COPY SHOPEON" en="" |
10:23:36 [offline-stt] 1.10s audio -> 115ms decode two-way -> vi (vi syllables 100% of 2, syllables) vi="KHÔNG ĐƯỢC" en="For example," |
10:23:36 [offline-bridge] #8 (endpoint) vi->en: "Không được" => "Not allowed" |
10:23:38 [offline-stt] 2.44s audio -> 221ms decode two-way -> vi (vi syllables 100% of 7, syllables) vi="ỦA ĐƯỢC HẢ CÓ VỤ ĐÓ HẢ" en="Or do have kid you don't have to do." |
10:23:39 [offline-bridge] #9 (endpoint) vi->en: "Ủa được hả có vụ đó hả" => "Huh, is that the case?" |
10:23:40 [offline-stt] 1.50s audio -> 160ms decode two-way -> vi (vi syllables 100% of 2, syllables) vi="TẠO MỚI" en="I think it's a good idea." |
10:23:40 [offline-bridge] #10 (endpoint) vi->en: "Tạo mới" => "Create new" |
10:23:45 [offline-stt] 3.76s audio -> 333ms decode two-way -> vi (vi syllables 100% of 6, syllables) vi="À ĐÂY LÀ ĐÂY LÀ XEM" en="I don't know. The second one is not." |
10:23:46 [offline-bridge] #11 (endpoint) vi->en: "À đây là đây là xem" => "Here it is, here it is, watch it" |
10:23:50 [offline-stt] 2.22s audio -> 218ms decode two-way -> vi (vi syllables 100% of 7, syllables) vi="ĐẤY CHÚNG TÔI CÓ HAI NGÔI RỪNG" en="That is not made up of a hand was in." |
10:23:51 [offline-bridge] #12 (endpoint) vi->en: "Đấy chúng tôi có hai ngôi rừng" => "Here we have two forests" |
10:23:52 [offline-stt] 1.98s audio -> 204ms decode two-way -> vi (vi syllables 100% of 2, syllables) vi="LĨNH VỰC" en="I did not push any other." |
10:23:52 [offline-bridge] #13 (endpoint) vi->en: "Lĩnh vực" => "Field" |
10:23:55 [offline-stt] 1.02s audio -> 120ms decode two-way -> vi (vi syllables 100% of 1, syllables) vi="Ừ" en="Boom." |
10:23:55 [offline-bridge] #14 (endpoint) vi->en: "Ừ" => "Yes" |
10:23:59 [offline-stt] 3.92s audio -> 333ms decode two-way -> vi (vi syllables 100% of 13, syllables) vi="NỘP CÁI NÀY ANH KIẾM CÁI THẰNG ĐÓ HỎI LUÔN BẢO HỒ NÈ" en="" |
10:24:00 [offline-bridge] #15 (endpoint) vi->en: "Nộp cái này anh kiếm cái thằng đó hỏi luôn bảo hồ nè" => "Submit this, then find that guy and ask him directly" |
10:24:01 [offline-stt] 1.38s audio -> 156ms decode two-way -> vi (vi syllables 100% of 2, syllables) vi="ĐƯỢC Ờ" en="Maybe" |
10:24:02 [offline-bridge] #16 (endpoint) vi->en: "Được ờ" => "Okay" |
10:24:08 [offline-stt] 1.82s audio -> 191ms decode two-way -> vi (vi syllables 100% of 2, syllables) vi="BÀ CON" en="" |
10:24:08 [offline-bridge] #17 (endpoint) vi->en: "Bà con" => "relatives" |
10:24:09 [offline-stt] 1.76s audio -> 178ms decode two-way -> vi (vi syllables 100% of 2, syllables) vi="MÌNH CÓ" en="Okay, do you wanna call it?" |
10:24:10 [offline-bridge] #18 (endpoint) vi->en: "Mình có" => "We have" |
10:24:12 [offline-stt] 1.60s audio -> 165ms decode two-way -> vi (vi syllables 100% of 1, syllables) vi="PHÒNG" en="" |
10:24:13 [offline-bridge] #19 (endpoint) vi->en: "Phòng" => "Room" |
10:24:17 [offline-stt] 3.62s audio -> 302ms decode two-way -> vi (vi syllables 100% of 6, syllables) vi="PHÂN MỘT CÁI BƯỚC SAU NÀY" en="There you go. Can you look at what's on it?" |
10:24:17 [offline-bridge] #20 (endpoint) vi->en: "Phân một cái bước sau này" => "Take one step next" |
10:24:22 [offline-stt] 5.60s audio -> 421ms decode two-way -> vi (vi syllables 100% of 21, syllables) vi="CÁI BÀI NÀY KHÔNG PHẢI LÀ CỨ NHÌN THẤY HAY LÀ LÀM THẾ NÀO ĐỂ NÓ TẠO RA ĐƯỢC CÁI NGUY" en="" |
10:24:24 [offline-bridge] #21 (endpoint) vi->en: "Cái bài này không phải là cứ nhìn thấy hay là làm thế nào để nó tạo ra được cái nguy" => "This article isn’t about simply seeing it or how to create harm from it" |
10:24:28 [offline-stt] 4.24s audio -> 356ms decode two-way -> vi (vi syllables 100% of 11, syllables) vi="THẾ NHƯNG CÓ CÁI NÀY VIẾT BÀI NÀY NỘP ĐÚNG KHÔNG" en="May let me copy my manual." |
10:24:29 [offline-bridge] #22 (endpoint) vi->en: "Thế nhưng có cái này viết bài này nộp đúng không" => "But is it okay to write and submit this article?" |
10:24:32 [offline-stt] 3.18s audio -> 271ms decode two-way -> vi (vi syllables 100% of 2, syllables) vi="THẾ NÀY" en="I mean," |
10:24:33 [offline-bridge] #23 (endpoint) vi->en: "Thế này" => "Like this" |
10:24:38 [offline-stt] 5.92s audio -> 449ms decode two-way -> vi (vi syllables 100% of 8, syllables) vi="ỨC ĐÚNG KHÔNG NGHĨA LÀ SAU NÀY Á" en="Uh yeah I'll go. Yeah, you can't sound good, yeah." |
10:24:39 [offline-bridge] #24 (endpoint) vi->en: "Ức đúng không nghĩa là sau này á" => "Yes, right, doesn't that mean in the future?" |
10:24:40 [offline-stt] 1.94s audio -> 183ms decode two-way -> vi (vi syllables 100% of 4, syllables) vi="SAU ĐÓ TỰ DO" en="But I think we might have to do it." |
10:24:40 [offline-bridge] #25 (endpoint) vi->en: "Sau đó tự do" => "Then freedom" |
10:24:42 [offline-stt] 2.52s audio -> 240ms decode two-way -> vi (vi syllables 100% of 5, syllables) vi="GIỜ TAO CỨ ĐIỀN HẾT" en="Yeah, they'll talk to you a day." |
10:24:43 [offline-bridge] #26 (endpoint) vi->en: "Giờ tao cứ điền hết" => "Now I just fill it all out" |
10:24:45 [offline-stt] 1.98s audio -> 203ms decode two-way -> vi (vi syllables 100% of 4, syllables) vi="THEO CÁI THÔNG TIN" en="Yeah, we talked in" |
10:24:46 [offline-bridge] #27 (endpoint) vi->en: "Theo cái thông tin" => "According to that information" |
10:24:47 [offline-stt] 1.48s audio -> 149ms decode two-way -> en (vi syllables 100% of 2, short-english) vi="CỦA BÀI" en="We do go by." |
10:24:47 [offline-bridge] #28 (endpoint) en->vi: "We do go by." => "Chúng ta tiếp tục đi thôi." |
10:24:49 [offline-stt] 2.64s audio -> 246ms decode two-way -> vi (vi syllables 100% of 6, syllables) vi="CỨ ĐIỂM XONG HẾT LÀ XONG" en="To the editorial, no." |
10:24:50 [offline-bridge] #29 (endpoint) vi->en: "Cứ điểm xong hết là xong" => "Once all the points are done, it's over." |
10:24:52 [offline-stt] 2.74s audio -> 247ms decode two-way -> en (vi syllables 0% of 1, syllables) vi="OK" en="Okay, so it's not it." |
10:24:53 [offline-bridge] #30 (endpoint) en->vi: "Okay, so it's not it." => "Được rồi, vậy thì không phải vậy." |
10:24:55 [offline-stt] 1.60s audio -> 160ms decode two-way -> en (vi syllables 0% of 0, syllables) vi="" en="And we'll just" |
10:24:56 [offline-bridge] #31 (endpoint) en->vi: "And we'll just" => "Và chúng ta chỉ" |
10:24:59 [offline-stt] 2.58s audio -> 237ms decode two-way -> en (vi syllables 67% of 6, syllables) vi="OLYN REBER LÀ GÌ Ý NHỞ" en="Opposite people" |
10:25:00 [offline-bridge] #32 (endpoint) en->vi: "Opposite people" => "Những người đối lập" |
10:25:02 [offline-stt] 2.18s audio -> 205ms decode two-way -> en (vi syllables 0% of 1, syllables) vi="WASHINGTON" en="We should say" |
10:25:03 [offline-bridge] #33 (endpoint) en->vi: "We should say" => "Chúng ta nên nói" |
10:25:06 [offline-stt] 1.92s audio -> 218ms decode two-way -> vi (vi syllables 100% of 7, syllables) vi="NÓI CHUNG THÌ KHÔNG PHẢI CÁI ĐẤY" en="Actually, we call my name." |
10:25:07 [offline-bridge] #34 (endpoint) vi->en: "Nói chung thì không phải cái đấy" => "In general, it’s not that thing" |
10:25:10 [offline-stt] 3.92s audio -> 370ms decode two-way -> vi (vi syllables 100% of 14, syllables) vi="THÌ ĐẤY LÀ CÁI KIỂU Ờ KIỂU CHƯA CÓ KẾT QUẢ HAY LÀ SAO" en="Today no thank you. Um so do you pick line is not?" |
10:25:11 [offline-bridge] #35 (endpoint) vi->en: "Thì đấy là cái kiểu ờ kiểu chưa có kết quả hay là sao" => "So is this a way of showing that there's no result yet?" |
10:25:12 [offline-stt] 2.30s audio -> 243ms decode two-way -> vi (vi syllables 100% of 2, syllables) vi="HAI HAI" en="" |
10:25:13 [offline-bridge] #36 (endpoint) vi->en: "Hai hai" => "Hai hai" |
10:25:15 [offline-stt] 2.24s audio -> 220ms decode two-way -> en (vi syllables 0% of 2, syllables) vi="REAL STOR" en="Standard source" |
10:25:16 [offline-bridge] #37 (endpoint) en->vi: "Standard source" => "Nguồn tiêu chuẩn" |
10:25:19 [offline-stt] 3.46s audio -> 312ms decode two-way -> en (vi syllables 0% of 1, syllables) vi="SHOP" en="Software maintains the bar." |
10:25:20 [offline-bridge] #38 (endpoint) en->vi: "Software maintains the bar." => "Phần mềm duy trì mức độ chất lượng." |
10:25:22 [offline-stt] 1.06s audio -> 129ms decode two-way -> vi (vi syllables 100% of 2, syllables) vi="CHÚNG TA" en="That's funny." |
10:25:23 [offline-bridge] #39 (endpoint) vi->en: "Chúng ta" => "We" |
10:25:26 [offline-stt] 4.06s audio -> 353ms decode two-way -> vi (vi syllables 90% of 10, syllables) vi="ỦA NHƯNG MÀ CÁI NÀY RỒI ADU ĐI REEN TA" en="" |
10:25:27 [offline-bridge] #40 (endpoint) vi->en: "Ủa nhưng mà cái này rồi adu đi reen ta" => "Hey, but this is already happening, right?" |
10:25:32 [offline-stt] 4.46s audio -> 376ms decode two-way -> en (vi syllables 50% of 2, syllables) vi="ONLINE Ở" en="So do you do online ambush?" |
10:25:34 [offline-bridge] #41 (endpoint) en->vi: "So do you do online ambush?" => "Vậy bạn có thực hiện các cuộc phục kích trực tuyến không?" |
10:25:37 [offline-stt] 2.52s audio -> 210ms decode two-way -> en (vi syllables 0% of 1, syllables) vi="ONLINE" en="Ola I mean" |
10:25:38 [offline-bridge] #42 (endpoint) en->vi: "Ola I mean" => "Ồ, tôi muốn nói là" |
10:25:47 [offline-stt] 4.42s audio -> 457ms decode two-way -> vi (vi syllables 94% of 16, syllables) vi="ỦA HÙNG QUANG CÔNG TY GÌ ẤY NHỜ QUANG NÓI NÓ VÔ LẠI ROTIC RỒI ANH" en="Oh, Tim Kong Tika, like Robotyro, and the other one." |
10:25:48 [offline-bridge] #43 (endpoint) vi->en: "Ủa hùng quang công ty gì ấy nhờ quang nói nó vô lại rotic rồi anh" => "Hey, what company is Quang? He said it’s terrible and bad, right?" |
10:25:51 [offline-stt] 2.56s audio -> 290ms decode two-way -> vi (vi syllables 100% of 8, syllables) vi="NÓ VẪN LÀM THÌ MỚI XẾP CÁI NÀY" en="So if you know what now you usually super name." |
10:25:52 [offline-bridge] #44 (endpoint) vi->en: "Nó vẫn làm thì mới xếp cái này" => "It will work only when this is arranged." |
10:25:53 [offline-stt] 1.62s audio -> 184ms decode two-way -> vi (vi syllables 100% of 4, syllables) vi="TƯỞNG NÓ NGHĨ RỒI" en="" |
10:25:53 [offline-bridge] #45 (endpoint) vi->en: "Tưởng nó nghĩ rồi" => "He thought he had figured it out" |
10:25:58 [offline-stt] 3.74s audio -> 342ms decode two-way -> vi (vi syllables 100% of 10, syllables) vi="CÁI CÁI CHẤT ẤY VẪN ĐƯỢC CHẠY LÊN DƯỚI NƯỚC" en="" |
10:25:59 [offline-bridge] #46 (endpoint) vi->en: "Cái cái chất ấy vẫn được chạy lên dưới nước" => "That substance still moves under water" |
10:26:04 [offline-stt] 5.22s audio -> 491ms decode two-way -> vi (vi syllables 94% of 16, syllables) vi="HEC MỚI ĐI ANH TẠI BÀI MÌNH CŨNG KHÁC MÀ HAI VOZON THÌ ĐÂU HỢP LÝ" en="Hunger mud money up, they my men on camera, tables and the colours." |
10:26:05 [offline-bridge] #47 (endpoint) vi->en: "Hec mới đi anh tại bài mình cũng khác mà hai vozon thì đâu hợp lý" => "Heck, your post is different, and two vozons don't make sense together" |
10:26:10 [offline-stt] 4.80s audio -> 429ms decode two-way -> vi (vi syllables 100% of 17, syllables) vi="Ờ KHÔNG Ý LÀ ĐÃ HỎI RỒI NÊN LÀ CỨ ĐIỆN THÔI TỰ MỌC VÀO NHAU ẤY" en="" |
10:26:11 [offline-bridge] #48 (endpoint) vi->en: "Ờ không ý là đã hỏi rồi nên là cứ điện thôi tự mọc vào nhau ấy" => "Oh, I already asked, so it just happens naturally between them." |
10:26:12 [offline-stt] 1.64s audio -> 186ms decode two-way -> vi (vi syllables 100% of 3, syllables) vi="Ờ ÁO THẾ" en="But without it?" |
End of preview. Expand in Data Studio
Cabin Offline — debug captures (vi ↔ en)
Real utterances recorded on a Galaxy S25 while testing the Cabin Offline two-way
interpreter (Menu → "Save utterance audio (debug)"), plus the logs and references used to
debug the on-device ASR / MT pipeline. Regression set for the cabin-offline-asr work.
587 utterances in 14 sessions (2026-10-19 → 2026-10-24), 16 kHz mono WAV.
| folder | content |
|---|---|
captures/<session>/NNN-<lang>.wav |
one utterance as the app cut it; <lang> = side the app routed it to |
captures/<session>/session.json |
source / target pair, two-way flag, start time, sample rate |
captures/<session>/replay/ |
desktop re-decode: asr.json, mt.json, Gemini Flash ref.json, report.md (3 sessions) |
captures-pro/<session>/ |
Gemini 2.5 Pro reference transcript + translation per utterance (3 sessions) |
online/2026-10-23-led-348b25/ |
15-min online (Soniox) on-stage session for comparison: audio parts, manifest, Gemini Pro/Flash refs |
logs/ |
phone logcat / card dumps from the test sessions |
metadata.jsonl |
one row per WAV: audio, session, utt_lang, source, target, and the Gemini Pro reference (transcript, translation) when one exists |
REGRESSION_NOTES.md |
what each session caught and how to rerun it |
References exist for 87 of the 587 utterances (sessions 2026-10-19T03-07-41Z,
2026-10-21T03-41-57Z, 2026-10-21T06-21-17Z). Some Flash ref.json entries contain the prompt
instead of a transcript (see notes); prefer captures-pro.
Published by the data owner, who confirmed the speakers' consent.
- Downloads last month
- 87