Yuanchao Li
@Li_Yuan_Chao
Followers
255
Following
843
Statuses
168
PhD candidate @EdinburghUni Doing #SpeechProc #NLProc #DigitalHealth #AffectiveComputing #HCI
Joined December 2014
#Interspeech2025 Please be sure to select “14.07 Responsible Speech Foundation Models (Special Session)” as your paper subject area when making a submission in the CMT system
📢📢CFP & Spread the word!! Again, "Responsible Speech Foundation Models" will appear as a special session @ISCAInterspeech #Interspeech2025 Last time, we got 8 great papers from CMU, CUHK, NTU, Apple, and more. Come join the party and win a paper award!!
0
0
1
@pengyf21 I sometimes ask speech researchers if they use ASR or SDS in their daily lives. The answers are pretty much the same: "No" with an awkward laugh😂
0
0
5
@jiatongshi My pleasure, Jiatong! As I am working on the acoustic similarity across audio, speech, and music, as well as their real and synthetic forms, I find VERSA to be a very timely and nice work. Bravo!
0
0
1
For those whose excellent work on speech foundation models was unfortunately not accepted by #ICASSP2025, you may consider resubmitting to our session at #Interspeech2025 😋😋
📢📢CFP & Spread the word!! Again, "Responsible Speech Foundation Models" will appear as a special session @ISCAInterspeech #Interspeech2025 Last time, we got 8 great papers from CMU, CUHK, NTU, Apple, and more. Come join the party and win a paper award!!
0
3
5
0
0
2
📢📢CFP & Spread the word!! Again, "Responsible Speech Foundation Models" will appear as a special session @ISCAInterspeech #Interspeech2025 Last time, we got 8 great papers from CMU, CUHK, NTU, Apple, and more. Come join the party and win a paper award!!
0
4
17
👏👏Check out the winner @enshi_zhang of our Generative Error Correction Challenge for LLM-based emotion recognition. They achieved 75.2% accuracy on ASR transcriptions of IEMOCAP using text refinement and GPT-4.0.
``Improving Speech-based Emotion Recognition with Contextual Utterance Analysis and LLMs,'' Enshi Zhang, Christian Poellabauer,
0
1
10
RT @ArxivSound: ``Large Language Model Based Generative Error Correction: A Challenge and Baselines for Speech Recognition, Speaker Tagging…
0
2
0
@Peppermint_2525 It’s too bad that Honda ended the ASIMO project just 3 months after I joined the Innovation Lab. Now the Innovation Lab has closed as well. Honda and several Japanese companies had the hardware resources but were just a few steps behind becoming Tesla. 体制のせいか
0
0
1
🙏🙏
惊闻日本知名声优大山羡代逝世，我馆深感痛惜。由大山羡代配音的《哆啦A梦》等经典作品为日中几代人的童年回忆。谨此表示由衷的哀悼。
0
0
0
👏👏Check out the 2nd best-performing system in our Generative Error Correction Challenge for LLM-based emotion recognition (yet unreasonably rejected by SLT2024). They achieved 75.1% accuracy on ASR transcriptions of IEMOCAP using GPT-4o.
``Context and System Fusion in Post-ASR Emotion Recognition with Large Language Models,'' Pavel Stepachev, Pinzhen Chen, Barry Haddow,
0
3
10
1
0
8
SIGDIAL was successfully hosted by Speech and Audio Processing lab @KyotoU_News where I did my master's study. Check out the keynote talk given by my senpai, Koji:
Successfully completed hosting SIGDIAL and YRRSDS this week, and also having my own keynote talk! Great to meet my new and old friends in Kyoto! Thanks for attending!
0
0
6
RT @shinjiw_at_cmu: Dear speech researchers, ICASSP'25 is looking for a reviewer in the speech and language areas. Please nominate yoursel…
0
15
0
Congrats to the award winners of our special session! @kalvin_chang @jiatong @OScharenborg @dmort27 @HungyiLee2 Since the organizers on-site were impressed by one of the presentations, and two papers received nearly the same number of votes, we decided to award more.
1
3
13
@OScharenborg You are right for regular papers, but in challenges, usually the top5 are invited to submit papers for acceptance. The papers I reviewed are the 1st and 2nd rankings. They also provided detailed description of the system and literature. There is no reason to reject them.
1
0
2
#icassp2025 new template is a mess. The default font of latex template is 10pt yet 9pt is acceptable. @ieeeICASSP also allows the author sec similar to past years. This means, 9pt+old author sec can give a whole extra page compared to 10pt+new author sec if there are many authors
2
7
18
@EmilianPostola1 seems no one likes the new template. We need to vote to change it back for next year🤔
1
0
2