Video yükleniyor...

Video Yüklenemedi

Ana Sayfaya Dön

New paper: Multi-Faceted Interactivity Alignment in Full-Duplex Speech Models We use RL to post-train speech models (Moshi and PersonaPlex) to talk more like a human: to know when to respond, when to wait, and when to nod along with “yeah”s and “okay”s when listening.

kyutai_labs's profile picture

kyutai

27,695 subscribers

33,244 görüntüleme • 2 ay önce •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 Yorum

Yorum bulunmuyor

Orijinal gönderinin yorumları burada görünecek

Benzer Videolar

We released Sonic-3.5 and Ink-2, the #1 streaming models for text to speech and speech to text you can use in your voice agents today. New architectures enable new frontiers for speed and quality. We're now the only provider to have #1 models for both speaking and listening.
2:47

We released Sonic-3.5 and Ink-2, the #1 streaming models for text to speech and speech to text you can use in your voice agents today. New architectures enable new frontiers for speed and quality. We're now the only provider to have #1 models for both speaking and listening.

Karan Goel

7,034,087 görüntüleme • 2 ay önce

How far are we from "Her"? I couldn't help but wonder, given the rapid fire of GPT-Live and Thinking Machines. Turns out, full duplex speech-to-speech is not particularly new. Moshi has been around for more than 2 years. Many thanks to Neil Zeghidour for bringing so much nuance. Check out kyutai who are doing god's work and open-sourcing it (not sponsored!). 00:00 Intro 01:24 ChatGPT Standard & LLM-based cascades 04:08 ChatGPT Advanced & end-to-end speech models 05:35 ChatGPT Live & Her 07:38 Thinking Machines & multi-stream modeling 09:13 Moshi & full duplex assistants 13:24 Moshi’s inner monologue 15:38 Why aren’t full duplex models everywhere? 17:08 Tool calling in full duplex models 18:00 Moshi RAG 19:06 Is the future full duplex? 20:24 How far are we from Her?
21:29

How far are we from "Her"? I couldn't help but wonder, given the rapid fire of GPT-Live and Thinking Machines. Turns out, full duplex speech-to-speech is not particularly new. Moshi has been around for more than 2 years. Many thanks to Neil Zeghidour for bringing so much nuance. Check out kyutai who are doing god's work and open-sourcing it (not sponsored!). 00:00 Intro 01:24 ChatGPT Standard & LLM-based cascades 04:08 ChatGPT Advanced & end-to-end speech models 05:35 ChatGPT Live & Her 07:38 Thinking Machines & multi-stream modeling 09:13 Moshi & full duplex assistants 13:24 Moshi’s inner monologue 15:38 Why aren’t full duplex models everywhere? 17:08 Tool calling in full duplex models 18:00 Moshi RAG 19:06 Is the future full duplex? 20:24 How far are we from Her?

Julia Turc

73,053 görüntüleme • 1 ay önce

When Bush came to NATO, he made a Speech and Left. When Obama came to NATO, he made a speech and left. When Trump came to NATO and he said “LET ME SEE YOUR BOOKS”
0:53

When Bush came to NATO, he made a Speech and Left. When Obama came to NATO, he made a speech and left. When Trump came to NATO and he said “LET ME SEE YOUR BOOKS”

MJTruthUltra

1,382,232 görüntüleme • 1 yıl önce

jungkook’s little nod and smile when he was listening to jimin talk about how he wanted to be in 2025 because bts would be together :(
0:25

jungkook’s little nod and smile when he was listening to jimin talk about how he wanted to be in 2025 because bts would be together :(

lex⁷

113,665 görüntüleme • 3 yıl önce

Speech to Speech is now available in 29 languages. In November, we launched Speech to Speech, enabling you to transform your voice into another character with full control over emotions, timing, and delivery. Today we're making Speech to Speech multilingual. Use Speech to Speech to get more control than with prompting alone. Simply say it how you want it, choose a target voice, and generate.
0:37

Speech to Speech is now available in 29 languages. In November, we launched Speech to Speech, enabling you to transform your voice into another character with full control over emotions, timing, and delivery. Today we're making Speech to Speech multilingual. Use Speech to Speech to get more control than with prompting alone. Simply say it how you want it, choose a target voice, and generate.

ElevenLabs

113,343 görüntüleme • 2 yıl önce

It's going to be insane when we can prompt our own virtual worlds like this and explore them. Imagine adding other characters you can talk to with LLMs + voice models!
0:28

It's going to be insane when we can prompt our own virtual worlds like this and explore them. Imagine adding other characters you can talk to with LLMs + voice models!

Justine Moore

2,189,886 görüntüleme • 1 yıl önce

Robert De Niro: I'm pretty close to being a free speech absolutist. Even speech I don't like. But when I hear something I don't like, I use my own free speech to respond. When I hear Trump say, "I don't think about Americans' financial situation." I say, "shut the fuck up."
0:42

Robert De Niro: I'm pretty close to being a free speech absolutist. Even speech I don't like. But when I hear something I don't like, I use my own free speech to respond. When I hear Trump say, "I don't think about Americans' financial situation." I say, "shut the fuck up."

Headquarters

797,562 görüntüleme • 2 ay önce

"We cannot remain silent. No-one can remain silent when there is an ongoing genocide and when people are denied the most basic human needs ... To stand with Palestine is to be human." Greta Thunberg calls out Israel's genocide during a speech in Leipzig, Germany.
0:55

"We cannot remain silent. No-one can remain silent when there is an ongoing genocide and when people are denied the most basic human needs ... To stand with Palestine is to be human." Greta Thunberg calls out Israel's genocide during a speech in Leipzig, Germany.

Quds News Network

728,916 görüntüleme • 2 yıl önce

Throwback to when iLLeY gave Dashy and Cellium a motivational speech 😂
0:39

Throwback to when iLLeY gave Dashy and Cellium a motivational speech 😂

SpawnUp

61,704 görüntüleme • 10 ay önce

ELON: WHEN WE LOSE FREE SPEECH, WE LOSE DEMOCRACY "Free speech is the bedrock of democracy, and free speech is what allows atrocities to be called out and for people to be aware of them. That's why the Germans wanted to shut down the press immediately. So, I think we should do everything we can to preserve free speech. And when we lose free speech, I think we lose democracy." Source: Interview with Ben Shapiro / Daily Wire, January 2024
0:23

ELON: WHEN WE LOSE FREE SPEECH, WE LOSE DEMOCRACY "Free speech is the bedrock of democracy, and free speech is what allows atrocities to be called out and for people to be aware of them. That's why the Germans wanted to shut down the press immediately. So, I think we should do everything we can to preserve free speech. And when we lose free speech, I think we lose democracy." Source: Interview with Ben Shapiro / Daily Wire, January 2024

Mario Nawfal

14,272,356 görüntüleme • 2 yıl önce

Speech-native models like Moshi sound great and answer fast, but aren’t as smart as text LLMs. In our new paper, MoshiRAG, we show how Moshi can ask for advice from a text LLM or a knowledge base. The tricky part is how to do this in real time without adding latency. 🧵
1:06

Speech-native models like Moshi sound great and answer fast, but aren’t as smart as text LLMs. In our new paper, MoshiRAG, we show how Moshi can ask for advice from a text LLM or a knowledge base. The tricky part is how to do this in real time without adding latency. 🧵

kyutai

53,142 görüntüleme • 4 ay önce

Listen up 🔊 We’ve made some updates to our Gemini Audio models and capabilities: — Gemini’s live speech-to-speech translation capability is rolling out in a beta experience to the Google Translate app, bringing you real-time audio translation that captures the nuance of human speech — Gemini 2.5 Flash and 2.5 Pro Text-to-Speech preview models bring improved adherence to style prompts, precision pacing with context-aware speed adjustments, and character voice consistency for multi-speaker scenarios — Gemini 2.5 Flash Native Audio is now updated, with improvements to handle complex workflows, navigate user instructions, and hold natural conversations
2:50

Listen up 🔊 We’ve made some updates to our Gemini Audio models and capabilities: — Gemini’s live speech-to-speech translation capability is rolling out in a beta experience to the Google Translate app, bringing you real-time audio translation that captures the nuance of human speech — Gemini 2.5 Flash and 2.5 Pro Text-to-Speech preview models bring improved adherence to style prompts, precision pacing with context-aware speed adjustments, and character voice consistency for multi-speaker scenarios — Gemini 2.5 Flash Native Audio is now updated, with improvements to handle complex workflows, navigate user instructions, and hold natural conversations

Google AI

1,093,874 görüntüleme • 8 ay önce

"Know when to walk away and know when to run"
0:23

"Know when to walk away and know when to run"

JScott Memes

57,182 görüntüleme • 11 ay önce

Our paper is out in Nature Human Behaviour🔥🔥 ‘Evidence of a predictive coding hierarchy in the human brain listening to speech’ 📄 💡Unlike language models, our brain makes distant & hierarchical predictions with Alexandre Gramfort and Jean-Rémi King Thread👇
0:43

Our paper is out in Nature Human Behaviour🔥🔥 ‘Evidence of a predictive coding hierarchy in the human brain listening to speech’ 📄 💡Unlike language models, our brain makes distant & hierarchical predictions with Alexandre Gramfort and Jean-Rémi King Thread👇

Charlotte Caucheteux @NeurIPS2025

145,183 görüntüleme • 3 yıl önce

Hiring RL Engineer! Started off as a curious project at Lossfunk to push the boundaries of LLMs in social reasoning - we are now building RL environments, data, and benchmarks to simulate more real-world scenarios. If you want to train SoTA RL models over multi-GPUs (H200s/B200s) to unlock next AI frontier, this is for you.
1:21

Hiring RL Engineer! Started off as a curious project at Lossfunk to push the boundaries of LLMs in social reasoning - we are now building RL environments, data, and benchmarks to simulate more real-world scenarios. If you want to train SoTA RL models over multi-GPUs (H200s/B200s) to unlock next AI frontier, this is for you.

Satpal Singh Rathore

46,010 görüntüleme • 1 yıl önce

"We know we're going to looks, know we're going to get chances. We just got to focus and bear down when they come." More post-game 🎥 »
0:29

"We know we're going to looks, know we're going to get chances. We just got to focus and bear down when they come." More post-game 🎥 »

Detroit Red Wings

19,832 görüntüleme • 5 ay önce

Tavus just secured a $40M raise to bring a new kind of intelligence to life. When technology starts to understand us, respond to us, and feel present with us, the future suddenly feels real. PALs add emotion, perception, and genuine human presence to computing in a way we’ve never seen before. Not a chatbot. Not a wrapper. A true human level interface you can talk to through video, voice, or text. This is what it looks like when machines finally learn to connect with us. Comment “PALs” and I’ll DM you free early access.
0:22

Tavus just secured a $40M raise to bring a new kind of intelligence to life. When technology starts to understand us, respond to us, and feel present with us, the future suddenly feels real. PALs add emotion, perception, and genuine human presence to computing in a way we’ve never seen before. Not a chatbot. Not a wrapper. A true human level interface you can talk to through video, voice, or text. This is what it looks like when machines finally learn to connect with us. Comment “PALs” and I’ll DM you free early access.

Manish Kumar Shah

19,375 görüntüleme • 9 ay önce

* abt love scenes 🐻 I will always go to him and say like it's okay we can do it. Bcs when when we’re filming love scene, we have to talk to eo first just want to feel free when we’re playing this scene. I want to both of us feel free and be happy playin it #ปอนด์ภูวินทร์
1:58

* abt love scenes 🐻 I will always go to him and say like it's okay we can do it. Bcs when when we’re filming love scene, we have to talk to eo first just want to feel free when we’re playing this scene. I want to both of us feel free and be happy playin it #ปอนด์ภูวินทร์

𝙇𝙚𝙖 🧡🤍

29,849 görüntüleme • 5 ay önce

Our latest speech-to-speech model is faster, more accurate, and excels at function calling. Watch @promptshant and Brian Fioca build a realtime voice agent that can search the web and hand off tasks to reasoning models with full context.
1:58

Our latest speech-to-speech model is faster, more accurate, and excels at function calling. Watch @promptshant and Brian Fioca build a realtime voice agent that can search the web and hand off tasks to reasoning models with full context.

OpenAI Developers

81,822 görüntüleme • 1 yıl önce