← All projects
GENERATIVE MODELS / 2026 — Present

Video-to-music generation

Exploring how video content can guide music generation, with an interest in extending the approach to longer videos.

Overview

Exploring how video content can guide music generation, with an interest in extending the approach to longer videos.

Approach

Reproducing a DiffRhythm-based baseline and investigating model improvements and evaluation for video-conditioned music generation.

My contribution

Leading the project, including baseline reproduction, experimental iteration, and the design of a long-video scoring framework.

Outcome

Ongoing experiments. Quantitative results and audiovisual examples will accompany a future update.