Functional MusicDIY Production, ResearchersMoûsaiAudio GenerationMoûsai is a diffusion model that can generate multiple minutes of high-quality stereo music at 48kHz from textual descriptions. It is also able to improve the quality of previously input audio samples.Website