Reproduce “SonoWorld4D: From One Video to a 4D Audio-Visual Scene via Diffusion Synchronization” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Derong Jin, Xiyi Chen, Ming C. Lin, Ruohan Gao
SonoWorld4D represents moving objects with persistent, object-attached audio tracks and synchronizes them with a video-conditioned camera-space mix, allowing audio to be rendered from new listener positions. At each diffusion step, it exchanges information between object- and camera-space audio. On SONOSCENE4D, its mean direction-of-arrival error is 40.7°, versus 55.5° for Sonic4D; its outdoor renderer omits reflections and reverberation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.