@jordan_dearsley @Vapi_AI Glad to see a better TTS eval coming! How to get a new model participating in this?
English
Bin Yang
21 posts
Bin Yang
@binyangderek
Founder & CEO @breezeblueX, building next-gen realtime interaction layer.
Toronto, Canada เข้าร่วม Temmuz 2014
598 กำลังติดตาม102 ผู้ติดตาม
You can feel whether the voice on a call is a human or a machine before you can explain why.
Today, @Vapi_AI is launching the Humanness Index™, a crowdsourced leaderboard for model humanness.
You are the benchmark. Cast your first vote today: humannessindex.vapi.ai
English
We've all been there, and we've all past that stage.
IMO, what needs to be optimized is "situational tts": generating speech that follows not just text, but role, intent, relationship, and scene. @BreezeBlueX is built for that purpose.
Neil Zeghidour@neilzegh
Our first commercial TTS model was optimized for WER and SSIM because that’s what research had taught us over years to be the standard metrics. The first customer feedbacks we had unveiled the huge blind spots of these metrics, in particular on naturalness, rhythm, emphasis, question intonation, etc. Now our internal eval has dozens of criteria monitored on each model.
English
@unilightwf fwiw, it's essentially a half-duplex model (speech in, text response out) with VAD integrated into the backbone as special tokens. They also defined some interesting interaction-related tasks (e.g., proactively respond to sounds).
English
Maybe a cool idea but really difficult for me to understand the main contribution (both from the paper and the demo)
arXiv Sound@ArxivSound
Zhifei Xie, Zihang Liu, Ze An, Xiaobin Hu, Yue Liao, Ziyang Ma, Dongchao Yang, Mingbao Lin, Deheng Ye, Shuicheng Yan, Chunyan Miao, "Audio Interaction Model," arxiv.org/abs/2606.05121
English
We definitely need more "real-world" benchmarks like this for speech models, if we want speech models to actually work in real life.
Bek@beknabdik
gpt-realtime-2 is genuinely fast, and this demo is great. no demo runs long enough to show the ceiling though. we ran it 60 turns. around 5 minutes in it went silent: took our audio, returned zero bytes, no error. then the connection dropped. the spec promises 128k tokens of context. the real ceiling is ~5 minutes. the previous gpt-realtime ran the same prompts comfortably. Full report here 🧵
English
We are brewing☕️ a turbo version⚡️ of Bluebell tts model for realtime applications.
Turns out, properly benchmarking end-to-end TTFA (time-to-first-audio, from client text -> tts provider server -> client intelligible speech) is nontrivial 👿
After we've sorted everything out (will open-source the benchmark tooling ofc), Bluebell-turbo-exp achieves the world's fastest TTFA 🥳
Guess who's the second one?
Bin Yang@binyangderek
~4 months in 2026 and we've seen 7 new voice design models released by different labs. Excited to see this new trend coming with @BreezeBlueX leading.
English
Agree that it's the behavior that defines full-duplex, not architecture. Also, I would highlight that "forecasting" is an important capability of full-duplex systems as it significantly reduces the "perceived latency".
Desh Raj@rdesh26
English
~4 months in 2026 and we've seen 7 new voice design models released by different labs. Excited to see this new trend coming with @BreezeBlueX leading.
English
What's intriguing about this benchmark is, text-based models (GPT-5.2) can achieve a pass rate of 85%, 2x that of the current best voice model. The gap is HUGE.
Logan Kilpatrick@OfficialLoganK
Our latest Live model is # 1 on Tau Voice Bench! Excited to see this new frontier of voice models cross the chasm of usability in production.
English
@DidiKieran Nice post! The follow-up work of VA-VAE, VTP (arxiv.org/abs/2512.13687) is also highly relevant, where the tokenizer is trained from scratch with a joint contrastive, self-supervised, and reconstruction objective to remove the dependency on pretrained representations.
English
Too many REPA / RAE / representation alignment papers lately?
I was lost too, so I wrote a blog post that organizes the space into phases and zooms in on what actually matters for general/molecular ML.
Curious what folks think - link below!
🔗 Blog: kdidi.netlify.app/blog/ml/2025-1…
English
@unilightwf from a tech perspective, human preference aligned TTS evaluation is hard. do you have any proposal here?
English
Open-source TTS界隈、ありがたいんだけど「高品質」「高速」と言いながらなんお評価結果も載せていなくて謎すぎる😭
日本語
@eigensteve Hi, could you share your filming setup? This should be the common practice for all online courses/tutorials.
English
First new video after being back from Sabbatical!!
PDE 101: Separation of Variables... or how I learned to stop worrying and solve Laplace's equation
One of the most important concepts in all of partial differential equations
youtube.com/watch?v=VjWtMl…
YouTube
English
Bin Yang รีทวีตแล้ว
Today, together with my collaborators Andreas Geiger, Philip Lenz and Christoph Stiller, I was awarded the 2021 Everingham Prize for KITTI at #ICCV21, which enabled many breakthroughs in #SelfDrivingCars. Thank you, truly an honor!
English
Check out our #ECCV2020 oral: Learning Lane Graph Representations for Motion Forecasting, which achieves best results on Argoverse benchmark.
Live QA: 8/27 9pm EDT
Poster session: 8/27 9am/7pm EDT
Paper: arxiv.org/abs/2007.13732
@RaquelUrtasun @UberATG
English