2 upvotes
Generative Flow Matching Splits Music Tracks
What is FLOMSS: Generative Music Source Separation by Flow Matching?
FLOMSS is a model that isolates individual instruments from a mixed audio recording using a generative flow-matching approach. It moves away from traditional discriminative separators by learning to reverse a diffusion-like process, yielding clearer instrument stems.
Who it is for
Audio engineers, remixers, and AI researchers can leverage the method for higher-quality source separation and better training data.