I'm an AI researcher, developer, and musician on a mission to teach AGI music. I believe artificial general intelligence should not only appreciate and understand music but also compose it.
- 🎵 YuE – an open and powerful music generation foundation model.
- 🎤 Leading music/audio research at the Multimodal Art Projection (MAP) Research Community.
- 🎓 I earned my Master's at CMU and am currently pursuing a PhD at HKUST.
- ⚡ Fun fact: I love pushing the boundaries of AI + music—so much that I sneak music-related problems into AI benchmarks, e.g. MMMU, OmniBench.
I'm always open to collaboration and new ideas. If you're excited about AI and music, join our Discord and create something amazing!
Potential topics:
- A publicly available Lyrics2Song dataset.
- Better diffusion upsampling for YuE – improving audio fidelity and resolution.
- An evaluation metric for musicality.
- Injecting more controls into YuE.
- A music arena for popular AI music systems.
- 🌍 NMLB (No Music Left Behind) – a large-scale project to collect ALL human music (not just western music) and build OPEN music understanding & generation models on top of it.