Renaissance Labs

People

Head of Research (co-founder) at Kyutai · Co-founder and Chief Science Officer at Gradium

Alexandre Défossez

Alexandre Défossez is an AI researcher in Paris working on generative music and speech models for low-latency use.

About

They co-founded Kyutai, where they lead research projects, and spent three years at FAIR Paris leading MusicGen and co-leading AudioCraft. They also co-founded the voice AI company Gradium.

Known for

  1. 01

    Moshi real-time spoken dialogue model and Mimi codec

  2. 02

    Demucs music source separation

  3. 03

    EnCodec neural audio codec

  4. 04

    MusicGen and AudioCraft at FAIR Paris

Activity

GitHub53 public contributions in the past year · 859 followers
1Dec2Jan11211Feb112MarApr122May13111JunJul272Aug2Sep1313Oct52Nov
Python
  • demucs★ 3.6k

    Code for the paper Hybrid Spectrogram and Waveform Source Separation

  • julius★ 462

    Fast PyTorch based DSP for audio and 1D signals

  • seewav★ 340

    Audio waveform visualisation, converts any audio to a nice video

  • Reproduction repository for the MDX 2021 Hybrid Demucs model

  • sdx23★ 30

    SDX23 startkit for the Demucs baselines.

Hugging Face11 models · 0 datasets · 0 spaces · 57 followers
2026: 1126

Repositories created per year

Research13 works · 1.4k citations · h-index 8 (OpenAlex)
2020: 1202023: 3232025: 3252026: 626

Works per year

Public data only, fetched 2026-11-17. GitHub counts cover public contributions; Hugging Face counts are repositories created; research data comes from OpenAlex.

Labs

Sources

Compiled from public sources. Something wrong, or would you like this page changed or removed? Tell us on Discord or open an issue on GitHub and we will act promptly.