KinWaiCheuk/source-separation

Installation

pip install -r requirements.txt

Reproducing results

python test.py checkpoint_path='./diffusion.ckpt'

Training Models

Training logs and checkpoints are saved inside outputs

Diffusion model

python train.py model=TConv128Diff

Encoder-Decoder model

python train.py model=TConv128

Unet model

python train.py model=UNet

Multi-modal loss

python train.py model=TConv128Spec