Text and/or audio to music. Trained on AudioSparx data, with opt-out for artists. Music is somewhat coherent, but struggles to produce anything with direction or character. Only tried text to music for now, could be a lot more interesting with audio prompts.
