PHD defence : Ninon Devis Salvy

  • Research
  • these

Ninon Devis Salvy, PhD candidate at ED 130 (Sorbonne University), audio researcher at Native Instruments, and sound artist specializing in artificial intelligence for music, will defend her dissertation entitled: “Creative Deep Learning on Real Time Embedded Architecture”, conducted within the Analyse-Synthèse team at IRCAM (UMR 9912 – STMS: CNRS, Sorbonne University, French Ministry of Culture), under the supervision of Charlotte Truchet and Carlos Agon.

Her work focuses on real-time neural synthesis, deep learning-based synthesizers, and generative model control. She is the creator of the Neurorack, the first Eurorack-format modular synthesizer powered by deep learning for audio generation. Through this work, she explores the intersection of creativity and artificial intelligence, using her own neural models to expand the expressive potential of musical performance.

The defense will be held in English, on Monday, May 12 at 3:00 PM, at IRCAM, in the Stravinsky Room.
It will also be streamed online at: https://youtube.com/live/4KoAmbZw6Xg

PhD Committee:

Cheng-Zhi Anna Huang, Associate Professor, MIT — Reviewer
Philippe Codognet, Professor, Sorbonne University — Reviewer
Joanna Demers, Professor, USC Thornton School of Music — Examiner
Laure Gonnord, Professor, Grenoble INP — Examiner
Anne Alombert, Lecturer, Paris 8 — Examiner
Pierre Saint-Germier, Research Scientist, CNRS — Examiner

Abstract:
Recent advances in deep learning have opened up new possibilities for real-time audio applications, particularly in music. This research explores the creation of a musical instrument that leverages these technologies to enable expressive, real-time sound generation and performance.

The dissertation introduces an innovative model for timbre manipulation based on perceptual audio features, enabling flexible transformations using architectures such as Fader Networks and RAVE. This work materializes in the Neurorack, a self-contained Eurorack-format synthesizer powered by deep learning, designed specifically for live use and creative exploration.
Finally, the thesis reflects on the role of artificial intelligence in creative processes, advocating for a design approach where the machine proposes without imposing—preserving artistic singularity while fostering a critical and meaningful dialogue between human and technology.

En poursuivant votre navigation sur ce site, vous acceptez l'utilisation de cookies pour nous permettre de mesurer l'audience, et pour vous permettre de partager du contenu via les boutons de partage de réseaux sociaux. En savoir plus.