Source codes for the DAFx2023 conference paper "Self-Supervised Disentanglement of Harmonic and Rhythmic Features in Music Audio Signals".
- Store the audio files uncer a certain path,
- In
configs/disentangle-2dvaegan-mono.yaml
, modifymodel.init_args.dset_config.path
to specify the audio path, - Run the following command.
python RunCLI.py fit --config configs/disentangle-2dvaegan-mono.yaml
An example code of audio generation is shown in Inference.ipynb
.