I actually found a Text-to-Audio model that is probably what Suno.ai bases their fine tuned generative music models on. https://github.com/suno-ai/bark