Open research toolkit extending MiniMax Music3 with arbitrary-audio encoding, continuation, inpainting, prepend generation, prompt-free generation, and automated evaluation.
#19
by coolpoodle - opened
Had to do a significant amount of work to get continuation, inpainting, and some Suno-esque modes working with MiniMax Music 3.
Caveat: it works, but the quality is not amazing yet.
That is exactly why I’m open-sourcing my research, code, experiments, and findings. I think the community can work together to push this into something much more capable and make Music 3 as awesome of an open model as possible.
Uploading the repo now!
Uploaded and done! https://huggingface.co/coolpoodle/music3lab
I will keep working on this but I hope the community also contributes 😃
Thanks guys.