Mamba-Diffusion Model with Learnable Wavelet for Controllable Symbolic Music Generation

For demonstrations, please visit our project website.

Training

python proffusion/main.py --model sdf_chd8bar --output_dir result/sdf_chd8bar

Inference

python proffusion/inference_sdf.py --chkpt_path=/path/to/checkpoint --uncond_scale=5.

Trained Checkpoints

We have released our pretrained checkpoint. Please download the checkpoint and put it under /result/ for inference.