Alexandre Défossez
Head of Research at Kyutai; former FAIR research scientist behind MusicGen, EnCodec, and AudioCraft
- Head of Research, Kyutai2
Alexandre Défossez is Head of Research at Kyutai, an open-science AI lab, and describes a research focus on multimodal language models. Before Kyutai, Défossez spent three years as a research scientist at FAIR at Meta, where Défossez led the music generation work behind MusicGen and co-led development of the AudioCraft framework. Défossez is first author of the EnCodec neural audio codec paper and of the Moshi speech-text foundation model paper; the Moshi repository releases its code under MIT and Apache licenses and its model weights under CC-BY 4.0.12345789
Work
AudioCraft (EnCodec and MusicGen) (external site: github.com)Open-source code
Co-led development of the AudioCraft framework, per Défossez's homepage. Défossez is first author of the EnCodec paper (2022) and a core-team author of the MusicGen paper (2023); the AudioCraft repository releases its code under the MIT license and its model weights under CC-BY-NC 4.0.13569
Moshi (external site: github.com)Open weights
First author of the Moshi paper (2024), a speech-text foundation model for real-time dialogue. The Moshi repository provides the code under MIT and Apache licenses and states that the model weights are released under CC-BY 4.0.78
Other documented roles
Roles the sources document but that are not confirmed as current.
- Research scientist at FAIR for three years, leading the music generation effort (MusicGen), Meta (FAIR)14
This profile covers public professional work only, from the sources below. It is not an endorsement or a ranking. To correct or remove it, use the correction form.