Kwanghee Choi

115 posts

Kwanghee Choi

Kwanghee Choi

@juice500ml

PhD student, working on speech AI with David Harwath @utsaltlab (@UTAustin) and David R. Mortensen @dmort27 (@LTIatCMU).

Katılım Ağustos 2023
165 Takip Edilen222 Takipçiler
Kwanghee Choi retweetledi
Ryan Soh-Eun Shim
Ryan Soh-Eun Shim@soheunshim·
✨New paper✨ We find script (e.g. Cyrillic, Latin) to be a linear direction in the activation space of Whisper, enabling transliteration at test-time by adding such script directions to the activations — producing e.g. Cyrillic Japanese transcriptions.
Ryan Soh-Eun Shim tweet media
English
1
1
3
216
Kwanghee Choi
Kwanghee Choi@juice500ml·
4 papers submitted & accepted at ACL 2026! 🎉 So grateful to work alongside & learn from amazing minds, pushing the boundaries of speech technologies, machine learning, and computational linguistics. See you in San Diego!
Kwanghee Choi tweet media
English
0
9
43
1.8K
Kwanghee Choi
Kwanghee Choi@juice500ml·
Thanks a lot for the interest in our work! Here's the recording for people who missed the seminar: youtu.be/DtFYKvNo9IQ
YouTube video
YouTube
Kwanghee Choi@juice500ml

𝐒𝐞𝐥𝐟-𝐬𝐮𝐩𝐞𝐫𝐯𝐢𝐬𝐞𝐝 𝐒𝐩𝐞𝐞𝐜𝐡 𝐌𝐨𝐝𝐞𝐥𝐬 𝐚𝐫𝐞 𝐏𝐡𝐨𝐧𝐨𝐥𝐨𝐠𝐢𝐜𝐚𝐥 𝐕𝐞𝐜𝐭𝐨𝐫 𝐌𝐚𝐜𝐡𝐢𝐧𝐞𝐬! 🗣️ Excited to be giving an invited talk this Thursday (March 19th, 3pm Amsterdam time)! Huge thanks to @mariannedhk at Univ. of Amsterdam for the invite 🙏

English
0
0
7
342
Kwanghee Choi
Kwanghee Choi@juice500ml·
Huge thanks for my wonderful coauthors, Eunjung and Cheol-jun, and my two favorite Davids, Mortensen 🐑 and Harwath 🤠 — best advisors I could ask for 🙏 Can't wait to see what we cook up next! 🚀
English
0
0
0
101
Kwanghee Choi
Kwanghee Choi@juice500ml·
🧵 Together, both papers take a step beyond the usual "what info do S3Ms encode" probing paradigm. We aim to answer how is that info actually encoded geometrically? Come see for yourself Thursday! 👀 Slides: docs.google.com/presentation/d…
English
1
0
1
114
Kwanghee Choi
Kwanghee Choi@juice500ml·
𝐒𝐞𝐥𝐟-𝐬𝐮𝐩𝐞𝐫𝐯𝐢𝐬𝐞𝐝 𝐒𝐩𝐞𝐞𝐜𝐡 𝐌𝐨𝐝𝐞𝐥𝐬 𝐚𝐫𝐞 𝐏𝐡𝐨𝐧𝐨𝐥𝐨𝐠𝐢𝐜𝐚𝐥 𝐕𝐞𝐜𝐭𝐨𝐫 𝐌𝐚𝐜𝐡𝐢𝐧𝐞𝐬! 🗣️ Excited to be giving an invited talk this Thursday (March 19th, 3pm Amsterdam time)! Huge thanks to @mariannedhk at Univ. of Amsterdam for the invite 🙏
Kwanghee Choi tweet media
English
1
5
17
2K
Kwanghee Choi retweetledi
Maarten Sap (he/him)
Maarten Sap (he/him)@MaartenSap·
🚀Apply to CMU LTI’s Summer 2026 “Language Technology for All” internship🎓Open to pre‑doctoral students new to language tech (non‑CS backgrounds welcome). 🔬12-14 weeks in‑person in Pittsburgh; travel + stipend paid.💸Deadline: Feb 20, 11:59pm ET. forms.gle/cUu8g6wb27HsWW…
English
8
88
599
84.3K
Kwanghee Choi retweetledi
Gallil Maimon
Gallil Maimon@GallilMaimon·
🎵💬 If you are interested in Audio Tokenisers, you should check out our new work! We empirically analysed existing tokenisers from every way - reconstruction, downstream, LMs and more. Grab yourself a ☕/🍺 and sit down for a read!
Gallil Maimon tweet media
English
1
25
103
5.9K
Kwanghee Choi retweetledi
Yifan Peng
Yifan Peng@pengyf21·
Excited to receive the Best Student Paper Award at #Interspeech2025 I started the OWSM project in 2023. It took me great effort to design a robust and scalable training framework using ESPnet, prepare unified data formats, and conduct large-scale training with academic resources.
Shinji Watanabe@shinjiw_at_cmu

Our work on OWSM v4 received the Best Student Paper Award at #Interspeech2025! 🏆🎉 Huge congratulations to the team! 🚀👏 I’m especially happy to see our open science efforts for speech foundation models recognized by the community. 🙌 🔗 isca-archive.org/interspeech_20…

English
0
5
35
2.1K
Kwanghee Choi retweetledi
William Chen
William Chen@chenwanch1·
We got the #Interspeech2025 Best Student Paper award! Congrats to @pengyf21, @m_shak33l, @yuisudo24, @MXzBFhjFpS1jyMI, and Chyi-Jiunn Lin!
William Chen tweet mediaWilliam Chen tweet mediaWilliam Chen tweet mediaWilliam Chen tweet media
Shinji Watanabe@shinjiw_at_cmu

Our work on OWSM v4 received the Best Student Paper Award at #Interspeech2025! 🏆🎉 Huge congratulations to the team! 🚀👏 I’m especially happy to see our open science efforts for speech foundation models recognized by the community. 🙌 🔗 isca-archive.org/interspeech_20…

English
2
10
53
3.9K
Kwanghee Choi retweetledi
Shinji Watanabe
Shinji Watanabe@shinjiw_at_cmu·
Our work on OWSM v4 received the Best Student Paper Award at #Interspeech2025! 🏆🎉 Huge congratulations to the team! 🚀👏 I’m especially happy to see our open science efforts for speech foundation models recognized by the community. 🙌 🔗 isca-archive.org/interspeech_20…
Shinji Watanabe tweet media
English
9
22
116
14K
Kwanghee Choi
Kwanghee Choi@juice500ml·
This wouldn't have been possible with my awesome co-first-author @mmiagshatoy and wonderful supervisors @shinjiw_at_cmu and Emma Strubell! I'll see you at Rotterdam, Wed 17:00-17:20 Area8-Oral4 (Streaming ASR)! (10/10)
English
0
0
2
127
Kwanghee Choi
Kwanghee Choi@juice500ml·
There's also bunch of engineering tricks that can improve the performance. We provide a pareto-optimal baseline after applying all the available tricks, positioning our work as a foundation for future works in this direction. github.com/Masao-Someki/S… (9/n)
Kwanghee Choi tweet media
English
1
0
3
116
Kwanghee Choi
Kwanghee Choi@juice500ml·
We also verified that DSUs are learnable with smaller weights (# of layers), i.e., more lightweight! This implies that we're using self-supervised models inefficiently when extracting DSUs. (8/n)
Kwanghee Choi tweet media
English
1
0
2
117
Kwanghee Choi
Kwanghee Choi@juice500ml·
We verified that DSUs are learnable with limited attention size (window size), i.e., streamable! This implies that DSUs are temporally "local". (7/n)
Kwanghee Choi tweet media
English
1
0
2
86
Kwanghee Choi
Kwanghee Choi@juice500ml·
After modifying the architecture, we fine-tune it with the DSUs extracted from the original full model. We're now understanding DSUs as "ground truth" for smaller models. (6/n)
Kwanghee Choi tweet media
English
1
0
2
95

Keşfet

@mariannedhk @pengyf21 @m_shak33l @yuisudo24 @MXzBFhjFpS1jyMI @WavLab @mmiagshatoy @shinjiw_at_cmu