C4DM Seminar: Côme Peladeau: Audio processors: estimating parameter distributions using discrete normalizing flows and DDSP
QMUL, School of Electronic Engineering and Computer Science
Centre for Digital Music Seminar Series
Seminar by: Côme Peladeau
Date/time: Wednesday, 22th October 2025, 1pm
Location: Online
Zoom Link: https://qmul-ac-uk.zoom.us/j/2387202947
Title: Audio processors: estimating parameter distributions using discrete normalizing flows and DDSP
Abstract: Audio processor, such as audio effects or synthesizers, are widely used in popular music production.
Their parameters control the quality of the output sound. Multiple combinations of parameters can lead to the same sound.
While recent approaches have been proposed to estimate these parameters given only the output sound, those are deterministic, i.e. they only estimate a single solution among the many possible parameter configurations.
This work proposes an approach to estimate all the combinations that lead to the target sound, in the form of a probability distribution.
We achieve this using differentiable signal processing and discrete normalizing flows.
Bio: Côme Peladeau graduated with a masters degree in Acoustics, Signal Processing and Computer Science applied to Music at Sorbonne Université in 2023.
He is now a PhD student working with Geoffroy Peeters and Dominique Fourer at the Information Processing and Communications Laboratory (LTCI) in Telecom Paris.
HI work focuses on audio effects estimation using deep learning and differentiable signal processing.