PeopleofSuperintelligence

Researcher

Neil Zeghidour

Neil Zeghidour is a speech and audio researcher turned founder who works on neural audio codecs, audio language models and real-time spoken dialogue. He studied at Paris-Dauphine University and the École normale supérieure, worked at Facebook's AI laboratory, and led generative audio work at Google DeepMind, where he was first author of the SoundStream codec in 2021 and a coauthor of AudioLM in 2022. His 2022 paper on learnable strides won an ICLR Outstanding Paper Award. In 2023, he co-founded the Kyutai lab, where he coauthored the Moshi voice model in 2024, and in 2025 he co-founded Gradium. He is CEO of Gradium, which extended its seed funding to $100 million in July 2026.

Now at
Kyutai and Gradium
Badges

Timeline

  1. By 2018

    Dauphine, ENS and Facebook AI

    He studies at Paris-Dauphine University and then at the École normale supérieure, and starts his career at Société Générale before joining Facebook's AI laboratory. In 2017 and 2018, he is first author of two papers on learning filterbanks from the raw waveform, with Emmanuel Dupoux, Nicolas Usunier and Gabriel Synnaeve among the coauthors. The 2018 paper reports the first end-to-end models trained from the raw signal to significantly outperform mel-filterbanks on a large vocabulary task.

  2. 2021

    SoundStream at Google

    At Google DeepMind, he starts and leads the generative audio team and builds the engineering teams behind neural audio codecs. In 2021, he is first author of SoundStream, a neural audio codec whose 3 kbps mode outperforms Opus at 12 kbps in subjective evaluations.

  3. 2022

    An ICLR Outstanding Paper

    His paper Learning Strides in Convolutional Neural Networks, written with Rachid Riad, Olivier Teboul and David Grangier, receives an Outstanding Paper Award at ICLR 2022. It introduces DiffStride, the first downsampling layer with learnable strides.

  4. 2022

    AudioLM for audio generation

    In 2022, he is one of eleven authors of AudioLM, a framework that maps audio to a sequence of discrete tokens and casts audio generation as a language modeling task.

  5. 2023

    Co-founding Kyutai

    He co-founds Kyutai in 2023, a non-profit AI research lab backed by €300 million from Xavier Niel, Rodolphe Saadé and Eric Schmidt. He is one of the five founding members of its scientific team.

  6. 2024

    Moshi, a real-time voice model

    In 2024, Kyutai releases Moshi, a speech-text foundation model for full-duplex dialogue with a theoretical latency of 160 ms and 200 ms in practice, and he is one of its eight authors. At Kyutai, he also leads the development of Hibiki, a real-time speech-to-speech translation model.

  7. 2025

    Co-founding Gradium

    In September 2025, he co-founds Gradium, a Paris-based voice AI startup, with Olivier Teboul, Laurent Mazaré and Alexandre Défossez. Gradium launches in December 2025 with a $70 million seed round led by FirstMark Capital and Eurazeo.

  8. As of October 2026

    CEO of Gradium

    He is co-founder and CEO of Gradium. In July 2026, the company extends its seed funding to $100 million, adds investors including Nvidia, and opens an office in the San Francisco Bay Area. He is also Audio Research Advisor at Kyutai.

Career

  1. Since 2025GradiumCo-founder and CEO
  2. Since 2023KyutaiCo-founder and founding scientific team member, Audio Research Advisor
  3. From 2021Google DeepMindLead, Generative Audio Team

Education

  1. Paris Dauphine University
  2. École Normale Supérieure

From the biography.

  1. 19% of people shaping AI who founded a dated organization have founded two or more, waiting a median of 7 years before a secondIncludes Neil Zeghidour. Across 1,136 people shaping AI who founded or co-founded a dated company, standalone lab or nonprofit, 221 (19%) founded at least two organizations, accounting for 36% of all 1,421 founding events.
  2. Google had the most people shaping AI every year from 2014 to 2025Includes Google. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  3. Google has the highest combined h-index among AI organizationsIncludes Google. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.
  4. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes Google. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Neil Zeghidour yet. Send us a free-licensed photo.

  1. Un Franco-Algérien dans le classement « des 50 qui vont faire demain » observalgerie.com. Checked 22 October 2026.
  2. Learning Filterbanks from Raw Speech for Phone Recognition arxiv.org. Checked 22 October 2026.
  3. End-to-End Speech Recognition From the Raw Waveform arxiv.org. Checked 22 October 2026.
  4. Neil Zeghidour, Slush 2026 speaker slush.org. Checked 22 October 2026.
  5. Meet Neil Zeghidour, speaker at HumanX humanx.co. Checked 22 October 2026.
  6. SoundStream: An End-to-End Neural Audio Codec arxiv.org. Checked 22 October 2026.
  7. ICLR 2022 Awards iclr.cc. Checked 22 October 2026.
  8. Announcing the ICLR 2022 Outstanding Paper Award Recipients blog.iclr.cc. Checked 22 October 2026.
  9. AudioLM: a Language Modeling Approach to Audio Generation arxiv.org. Checked 22 October 2026.
  10. Gradium, premier spin-off de Kyutai, lève 60 millions d'euros pour devenir une référence mondiale de l'IA vocale maddyness.com. Checked 22 October 2026.
  11. Moshi: a speech-text foundation model for real-time dialogue arxiv.org. Checked 22 October 2026.
  12. Gradium: Solving voice gradium.ai. Checked 22 October 2026.
  13. Gradium secures USD $70 million to launch AI voice platform cfotech.com.au. Checked 22 October 2026.
  14. Nvidia Joins $100 Million Bet on Gradium Voice AI pymnts.com. Checked 22 October 2026.
  15. Kyutai Team kyutai.org. Checked 22 October 2026.