Comment by acacac

will the model ever be extended to being able to segment audio (eg. different people talking, different instruments in a soundtrack?)