This is a simple and controllable music generation model that will generate 12 seconds of audio from instructions you provide. You can optionally provide reference audio from which to extract a wide range of melodies. The model will then try to follow the provided description and melody.
Visit Official Website