Shih-Lun (Sean) Wu

Hi, there! I’m Shih-Lun. I am a Ph.D. student in CSAIL & EECS at Massachusetts Institute of Technology (MIT). I am fortunate to be advised by Prof. Anna Huang, and we strive to push the frontiers of music generation & interactions, controllable generative models, and preference alignment/tuning. Recently, we built MIDI-LLM (with live demo!) by adapting Llama LLM for text-to-MIDI music generation.

I graduated in 2024 with a M.Sc. in Language Technologies from CMU’s Language Technologies Institute, School of Computer Science, where I was advised by Prof. Shinji Watanabe and Prof. Chris Donahue. I worked on controllable music generation, audio captioning, and spoken language understanding. I was also a two-time research scientist intern at Adobe Research (mentor: Dr. Nick Bryan), where we built Stemphonic (2025) and Music ControlNet (2023).

Before CMU, I received my B.Sc. degree (in Computer Science) from National Taiwan University. Also, I’ve been with two vibrant Taiwanese AI R&D teams: Asus AICS Center, and Taiwan AI Labs, working as a software dev intern first, and later as a research engineer.

My undergraduate research focused on symbolic-domain music generation, where I was advised by the wonderful Dr. Yi-Hsuan Yang. Feel free to listen to our model’s creative works here, or even compose with it! I’ve also worked with Prof. Chung-Wei Lin and Prof. Eunsuk Kang on formal verification under weakly-hard constraints.