A new music gen model. Still in training. Model weights are in the HF repo's branches/revisions.

Audio samples:

After ~8 hours of training:

<audio controls src="https://cdn-uploads.huggingface.co/production/uploads/62e54f0eae9d3f10acb95cb9/WHj8dXbgFU-14FOz6vbal.wav"></audio>

Not great yet, but at least it learned to follow lyrics