Yinghao Ma
Yinghao Ma is currently studying for a PhD in Computer Science and is a member of C4DM. He was recently awarded a Google Fellowship in Machine Perception.
What’s your thesis title?
My thesis was originally titled "SSL4Music" but has recently evolved into "LLMs for Music Understanding and Generation" to reflect the shift toward large language models for multimodal music understanding and generation.
Can you summarise your research in one sentence?
I am developping large language models that can understand, generate, and interact with music in a human-aligned way.
Why did you choose to do your doctoral research at QMUL?
Queen Mary offers one of the strongest platforms in the world for AI and music research. My supervisor, Dr. Emmanouil Benetos, is an ideal match for my research interests—highly knowledgeable, supportive, and open-minded. QMUL also provides a unique environment that connects music research with cutting-edge work in NLP, multimodal AI, and industry applications globally, which is exactly where I believe the future of music technology lies.
How does your research group support you?
The group provides both technical and musical support—ranging from engineering advice and music knowledge help to guidance on traditional music information retrieval (MIR) methods and research feedback. I benefit from a collaborative environment where people are willing to share expertise and challenge ideas constructively in a boarder music AI research field.
What’s a typical research day like for you?
I usually start by reading newly released papers or reviewing papers assigned to me, then continue writing survey sections or proposals. Most days involve meetings with collaborators, discussing code, and debugging model or training issues. Research for me is a continuous mix of exploration, implementation, and communication.
What’s been your most exciting research experience so far?
Seeing my open-source model MERT reach over 10,000 monthly downloads for more than three consecutive years has been incredibly rewarding. It showed me that well-designed research can have long-term impact in both academia and industry.
Any advice for anyone about to start their PhD journey?
Become a strong problem-solver—especially with engineering and infrastructure issues. Stay open to emerging technologies, and don’t isolate yourself: actively collaborate, talk to industry, and learn from researchers in neighbouring fields. A PhD is not just about depth, but also about building a broad and adaptive mindset.