Ben-Yee/Music_Source_Separator

Music source separation tools with pytorch. Train, separate, evaluate on custom datasets with TF and time domain models.

Music Source Separator

Models Available:

ConvTasNet, Spleeter, D3Net, DCCRN (updating)

How to use:

1. Add datasets to the dataset folder

2. Configure the General_config.py for direrctory setup, and configure the Model_config.py for model parameters

3. python3 train.py / python3 separate.py / python3 eval.py

References: https://github.com/deezer/spleeter https://github.com/tky823/DNN-based_source_separation https://github.com/generalwave/spleeter.pytorch