/DeepSeparationClustering

Using Deep Clustering Algorithm for monaural source separation

Primary LanguagePythonMIT LicenseMIT

Deep Clustering for source separation

Experimenting with Deep Clustering Algorithm for monaural source separation.

Usage

In configuration.py, set data_dir to folder containing test files and results_dir to output folder and run test.py

Training

run train.py for training on entire train folder in data_dir

Testing and visualization

Only track wise testing as of now run test.py followed by visualization.py

Dependencies

  • python 3.x,
  • keras2.x
  • SciPy
  • pysoundfile
  • scikit-learn
  • matplotlib (for visualization)

Reference

  1. Deep clustering: Discriminative embeddings for segmentation and separation