Neil Zeghidour

Research at Google

As a research scientist at Google Research, Zeghidour worked on learned audio representations and generation. Zeghidour was first author of LEAF, a learnable frontend for audio classification posted in January 2021, and of SoundStream, an end-to-end neural audio codec that the Google Research blog introduced later that year.⁹⁴³

Zeghidour was also the last author of AudioLM, which treats audio generation as a language-modeling task over discrete audio tokens, and co-wrote the 2022 Google Research blog post presenting it with Zalán Borsos.⁶⁵

Kyutai

Kyutai's team page lists Zeghidour as Audio Research Advisor. Zeghidour is the last author of the 2024 Moshi paper, which describes a speech-text foundation model for real-time spoken dialogue, and of the 2025 paper on Hibiki, a model for high-fidelity simultaneous speech-to-speech translation. Zeghidour's GitHub profile pins the kyutai-labs repositories for Moshi and Hibiki.¹⁷¹⁰²

Work and contributions

• SoundStream Published research

  First author of SoundStream: An End-to-End Neural Audio Codec (2021) and co-author of the Google Research blog post introducing it, which credits the work to Zeghidour, Alejandro Luebs, Ahmed Omran, Jan Skoglund, and Marco Tagliasacchi.⁴³

• AudioLM Published research

  Last author of AudioLM: a Language Modeling Approach to Audio Generation (2022) and co-author, with Zalán Borsos, of the Google Research blog post presenting it.⁶⁵

• Moshi Published research

  Co-author of Moshi: a speech-text foundation model for real-time dialogue (2024), with Alexandre Défossez and other Kyutai researchers.⁷⁸

Timeline

1. Jan 21, 2021Posted LEAF: A Learnable Frontend for Audio Classification as first author.⁹
2. Jul 7, 2021Posted SoundStream: An End-to-End Neural Audio Codec as first author.⁴
3. Sep 7, 2022Co-authored AudioLM: a Language Modeling Approach to Audio Generation, as last author.⁶
4. Sep 17, 2024Co-authored the Moshi speech-text foundation model paper from Kyutai, as last author.⁷
5. Feb 5, 2025Co-authored High-Fidelity Simultaneous Speech-To-Speech Translation, the Hibiki paper, as last author.¹⁰

Other documented roles

Roles the sources document that are not confirmed as current.

• Research Scientist, Google Research (blog bylines, 2021 and 2022), Google Research³⁵

Where they work

• Kyutai 2 people in the wiki

Related people

• Alexandre DéfossezAlso at Kyutai
• Jonathan ShenSpeech and audio · Image, video, and 3D generation
• Sanyuan ChenSpeech and audio · Image, video, and 3D generation
• Ju-Chieh ChouSpeech and audio · Image, video, and 3D generation
• Xu TanSpeech and audio · Image, video, and 3D generation
• Prafulla DhariwalImage, video, and 3D generation · Speech and audio
• Jade CopetImage, video, and 3D generation · Speech and audio
• Heiga ZenSpeech and audio · Image, video, and 3D generation
• Yi RenSpeech and audio · Image, video, and 3D generation
• RJ Skerry-RyanSpeech and audio · Image, video, and 3D generation