International Symposium on Diffusion Models for Audio and Music Processing - 2025
The International Symposium on Diffusion Models for Audio and Music Processing was held on 15–16 November 2025 at Osnabrück University.
The symposium explored diffusion models, a state-of-the-art deep learning technique used to create high-quality and highly realistic audio and music content. Leading researchers presented their latest work and shared their insights, providing an excellent opportunity for specialized knowledge exchange and networking.
Organizer: Dr. Maurício do Vale Madeira da Costa
Date: 15-16 November
Videos of all presentations are available on YouTube
Technical Program:
| Date | Time | Name | Title | Affiliation |
| Saturday (Nov. 15th) | 13:00 | Registration | ||
| 13:30 | Dr. Jean-Marie Lemercier | Diffusion-based Techniques for Speech Restoration | NVIDIA | |
| 14:30 | Dr. Eloi Moliner Juanpere | Diffusion Models for Audio Effect Estimation and Automatic Music Mixing | Aalto University, Acoustics Lab | |
| 15:30 | Coffeebreak | |||
| 16:00 | Dr. Javier Nistal | Context-aware Music Stem Co-creation via Flow Matching | Sony CSL | |
| 17:00 | Dr. Michele Mancusi | One Model, Many Tricks: Bending Diffusion Models at Inference Time | Music.AI / Moises | |
| Sunday (Nov. 16th) | 09:00 | Dr. Ben Maman | Multi-Aspect Conditioning for Diffusion-Based Music Synthesis | International Audio Laboratories Erlangen |
| 10:00 | Teysir Baoueb and Bernardo V. de Miranda | Conditional Diffusion Model for Audio Generation | Télécom Paris / ADASP, PhD candidate | |
| 11:00 | Dr. Maurício do V. M. da Costa | High-Definition Time-Frequency Representations: Applications on Diffusion Models | Universität Osnabrück / MTDML, postdoctoral researcher | |
| 12:00 | Round Table | |||