🔥Telegram İfşa

I've been working on a custom animation system in Unity for a couple months now, and thought I'd show off a bit of the progress! The system focuses on enabling modularity and "complex" compositing of animation data in an easy to use (and extend) package. I just need to prefice

Talk to unmute.sh 🔊, the most modular voice AI around. Empower any text LLM with voice, instantly, by wrapping it with our new speech-to-text and text-to-speech. Any personality, any voice. Interruptible, smart turn-taking. We’ll open-source everything within the

🚀 Say hello to unmute.sh — a modular voice AI system built on our in-house low latency text-to-speech and speech-to-text engines. It works in English🇬🇧and French 🇫🇷 and you can customize the voice and personality 🎙️Try it live and tell us what you think!

Unmute.sh is out! I hope you have as much fun with it as we did during development.

We just released unmute.sh 🔇🔊 It is a text LLM wrapper, based on in-house streaming ASR, TTS, semantic VAD to reduce latency⏱️ Unlike Moshi 🟢, Unmute 🔊 is turn base, but allows customization in two clicks🖱️: voice and prompt! Paper and open source coming soon.

Unmute is our new cascaded voice assistant: fast, accurate, and flexible. It doesn't have the full-duplex and zero latency of Moshi, but you can change the voice with a 10s sample and plug any LLM. A good playground for testing custom voice AIs.

Kyutai Speech-To-Text is now open-source! It’s streaming, supports batched inference, and runs blazingly fast: perfect for interactive applications. Check out the details here: kyutai.org/next/stt

Kyutai TTS and Unmute are now open source! The text-to-speech is natural, customizable, and fast: it can serve 32 users with a 350ms latency on a single L40S. Try it out and get started on the project page: kyutai.org/next/tts

Mistral is proud to provide the text LLM powering Unmute, the open-source voice AI from kyutai!

🎮 1/ India stands at the brink of a cultural renaissance globally through game-first IPs which will surpass the impact of Bollywood. This is what I cover in Part 2 of my series on Cultural Diplomacy through AAA Gaming substack.com/@ngranati/note…

I’m happy to share that I’ll be attending ICML 2025 in Vancouver next week to present 𝐇𝐢𝐛𝐢𝐤𝐢 [github.com/kyutai-labs/hi…] 🇫🇷🇬🇧 — Kyutai’s real-time and expressive speech translation system. I'll be presenting the poster on Wednesday, July 16 at 4:30PM, feel free to stop by! 💬

If you're at #ICML2025 this week, come check out these 3 posters from our lab🟢! - Aligning Spoken Dialog Models from User Interactions, Anne Wu Thu 17 Jul 11am-1:30pm W-316 - High-Fidelity Simultaneous Speech-To-Speech Translation, Tom Labiausse Wed 16 Jul 4:30pm - 7pm

Introducing General Intuition and our $133.7M Seed from Khosla Ventures, General Catalyst, and Raine. We build foundation models and general agents for environments that require deep spatial and temporal reasoning.

🚀New models: ARC-Encoders We introduce a lightweight encoder that compresses context into continuous representations for LLMs, reducing inference cost while preserving performance. Our Adaptable text Representations Compressor, named ARC-Encoder, achieves large efficiency gains

1/2 We’re releasing an in-depth tutorial on neural audio codecs, the secret sauce that makes it possible for audio LLMs to not sound like a horror movie:

My article is out! It was inspired by Andrej Karpathy's "RNN effectiveness" blog post that got me into ML around 10 years ago. Wacky samples and fancy animations galore

My blog post just topped HackerNews!! 😍

🚨 Wanna join a laser-focused AI research lab in Europe? Kyutai is hiring! Based in central Paris, we are a highly-funded non-profit lab where daring open research meets the real world, paving the way to tomorrow's technologies. It is all about focus, passion and speed. With no

Text-to-Image models don't need 3 training stages anymore! 🤯 Our new MIRO method integrates human alignment directly into pretraining. 19x faster convergence ⚡ 370x less compute than FLUX-dev 📉 Train once, align to many rewards. The era of multi-stage training is over!

Training a quantized autoencoder for Fashion MNIST - an animation from our new tutorial on neural audio codecs. The two might seem unrelated, but quantization is a key building block of these codecs.