1 paper across 1 session
We propose unified latent diffusion framework for simultaneous music generation, source imputation, and query-driven arbitrary source extraction.
