MR-FlowDPO: Multi-Reward Direct Preference Optimization for Flow-Matching Text-to-Music Generation
ICASSP 2026
Alon Ziv
I'm a computer science PhD student at the Hebrew University of Jerusalem at the SLP Lab, and a research engineer at Google DeepMind working on Producer.ai.
My research interests are in fundamental AI models for audio. I currently study methods for real-time, controllable music generation. I am fortunate to be advised by Dr. Yossi Adi.
Previously, I've been a research scientist intern at FAIR Meta AI (audio generation) and a machine learning researcher at Mobileye (autonomous vehicle perception).
I love music!!!
In parallel with computer science, I'm also a semi-professional musician. I play piano and drums, and I have some experience in music production.
alonzi at cs.huji.ac.il
News
- [February 2026] Joined Google DeepMind as a Research Engineer.
- [January 2026] MR-FlowDPO paper was accepted to ICASSP 2026.
- [May 2025] We launched Vibes 1.1 at Riffusion.
- [March 2025] We launched Vibes 0.8 at Riffusion.
- [February 2025] Joined Riffusion (Producer.ai) as a Research Scientist.
- [January 2025] We open-sourced JASCO as part of the AudioCraft library.
- [November 2024] Recieved a Best Reviewer Award at ISMIR 2024.
- [October 2024] Our Mobileye Patent on LiDAR-Camera fusion was published.
- [September 2024] Research internship at FAIR NYC, Audio Generation.
- [June 2024] JASCO paper was accepted to ISMIR 2024.
- [May 2024] Research internship at Riffusion: Vibes.
- [April 2024] Our Mobileye Patent on Vehicle Navigation based on Aligned Image and LiDAR Information was published.
- [January 2024] MAGNeT paper was accepted to ICLR 2024.
- [January 2024] We open-sourced MAGNeT as part of the AudioCraft library.
- [June 2023] Research internship at FAIR Tel Aviv, Audio Generation.
- [February 2023] Our Mobileye Patent on LiDAR and REM Localization was published.
- [October 2022] Started my PhD at The Hebrew University, Advisor: Dr. Yossi Adi.
Publications
