Skip to content

@kyutai-labs kyutai

Change the repository type filter

All

    Repositories list

    28 repositories

    • moshi

      Public
      Moshi is a speech-text foundation model and full-duplex spoken dialogue framework. It uses Mimi, a state-of-the-art streaming neural audio codec.
      Python
      •
      Apache License 2.0
      •956•10k•68•16•Updated May 16, 2026May 16, 2026

    • pocket-tts

      Public
      A TTS that fits in your CPU (and pocket)
      Python
      •
      MIT License
      •509•4.5k•42•7•Updated May 5, 2026May 5, 2026

    • tts_longeval

      Public
      Python
      •
      MIT License
      •1•30•0•0•Updated Apr 29, 2026Apr 29, 2026

    • moshi-rag

      Public
      MoshiRAG is a compact full-duplex speech language model augmented with asynchronous knowledge retrieval to improve factuality without sacrificing real-time inte…
      Rust
      •
      Apache License 2.0
      •6•87•0•0•Updated Apr 28, 2026Apr 28, 2026

    • flashy

      Public
      Framework for writing deep learning training loops. Lightweight, and retaining full freedom to design as you see fits. It handles checkpointing, logging, distri…
      Python
      •
      MIT License
      •0•6•0•0•Updated Apr 24, 2026Apr 24, 2026

    • invincible-voice

      Public
      To bring back voice to those who lost it
      TypeScript
      •
      MIT License
      •7•91•6•1•Updated Apr 20, 2026Apr 20, 2026

    • ovie

      Public
      Official implementation and models for OVIE (One View Is Enough! Monocular Training for In-the-Wild Novel View Generation)
      Jupyter Notebook
      •3•66•0•0•Updated Apr 16, 2026Apr 16, 2026

    • dactory

      Public
      Python
      •
      Apache License 2.0
      •5•52•0•0•Updated Apr 2, 2026Apr 2, 2026

    • unmute

      Public
      Make text LLMs listen and speak
      Python
      •
      MIT License
      •227•1.3k•26•1•Updated Mar 26, 2026Mar 26, 2026

    • flash-attn3-jax

      Public
      JAX bindings for the FlashAttention 3 kernels
      C++
      •
      BSD 3-Clause "New" or "Revised" License
      •1•23•0•0•Updated Mar 9, 2026Mar 9, 2026

    • casa

      Public
      A vision-language model with an improved cross-attention mechanism for scalable streaming inference
      Python
      •
      MIT License
      •3•29•3•0•Updated Mar 9, 2026Mar 9, 2026

    • yomikomi

      Public
      A small rust-based data loader
      Rust
      •
      Apache License 2.0
      •2•37•1•1•Updated Feb 20, 2026Feb 20, 2026

    • hibiki-zero

      Public
      A real-time and multilingual speech translation model
      Python
      •
      MIT License
      •24•248•2•0•Updated Feb 13, 2026Feb 13, 2026

    • delayed-streams-modeling

      Public
      Kyutai's Speech-To-Text and Text-To-Speech models based on the Delayed Streams Modeling framework.
      Python
      •
      Apache License 2.0
      •306•2.9k•35•0•Updated Jan 26, 2026Jan 26, 2026

    • dora

      Public
      Dora is an experiment management framework. It expresses grid searches as pure python files as part of your repo. It identifies experiments with a unique hash s…
      Python
      •
      MIT License
      •0•5•0•0•Updated Jan 22, 2026Jan 22, 2026

    • sphn

      Public
      python bindings for symphonia/opus - read various audio formats from python and write opus files
      Rust
      •
      Apache License 2.0
      •9•79•1•0•Updated Jan 7, 2026Jan 7, 2026

    • ARC-Encoder

      Public
      Python
      •
      Apache License 2.0
      •3•28•0•0•Updated Jan 5, 2026Jan 5, 2026

    • jax-flash-attn3

      Public
      JAX bindings for the flash-attention3 kernels
      C++
      •3•22•0•0•Updated Jan 2, 2026Jan 2, 2026

    • neural-audio-codecs-anims

      Public
      Animations for the blog "Neural audio codecs: how to get audio into LLMs"
      TypeScript
      •0•4•0•0•Updated Oct 20, 2025Oct 20, 2025

    • nanoGPTaudio

      Public
      Code for the blog "Neural audio codecs: how to get audio into LLMs"
      Python
      •
      MIT License
      •10k•167•0•0•Updated Oct 20, 2025Oct 20, 2025

    • moshi-finetune

      Public
      Python
      •
      Apache License 2.0
      •64•451•14•1•Updated Oct 3, 2025Oct 3, 2025

    • moshi-swift

      Public
      Swift
      •
      MIT License
      •16•137•2•0•Updated Jun 26, 2025Jun 26, 2025

    • hibiki

      Public
      Hibiki is a model for streaming speech translation (also known as simultaneous translation). Unlike offline translation—where one waits for the end of the sourc…
      Rust
      •
      Apache License 2.0
      •117•1.5k•9•1•Updated Apr 15, 2025Apr 15, 2025

    • moshivis

      Public
      Kyutai with an "eye"
      Python
      •
      Apache License 2.0
      •32•248•1•0•Updated Mar 26, 2025Mar 26, 2025

    • kaudio

      Public
      Rust crate for some audio utilities
      Rust
      •
      Apache License 2.0
      •0•28•0•0•Updated Mar 8, 2025Mar 8, 2025

    • moshi-webrtc

      Public
      Proof of concept for running moshi/hibiki using webrtc
      Rust
      •
      Apache License 2.0
      •2•21•0•0•Updated Feb 28, 2025Feb 28, 2025

    • jax-flash-attn2

      Public
      JAX bindings for the flash-attention2 kernels
      C++
      •0•12•0•0•Updated Jan 16, 2025Jan 16, 2025

    • ogg-table

      Public
      Ogg-vorbis reader with fast random access
      Rust
      •
      Other
      •1•8•0•0•Updated Aug 29, 2024Aug 29, 2024
    ProTip! When viewing an organization's repositories, you can use the props. filter to filter by custom property.