nussl (pronounced nuzzle) 1 is a flexible, object oriented python audio source separation library created by the Interactive Audio Lab at Northwestern University. At its core, nussl provides implementations of common source separation algorithms as well as an easy-to-use framework for prototyping and adding new algorithms. The aim of nussl is to create a low barrier to entry for using popular source separation algorithms, while also allowing the user fine tuned control of low-level parameters.

First steps

• Getting Started
  • Installation
    • Requirements
    • pip install
    • Running the tests
    • Installing from source and downloading
    • Troubleshooting
      • Matplotlib issues
      • Contact
  • Introduction to AudioSignals
    • AudioSignal Basics
      • Initialization from a file
      • Initialization from a numpy array
      • Other basic manipulations
    • Time-Frequency Representations
      • STFT Basics
      • Inverse STFTs
      • STFT Parameters
  • Separation via Time-Frequency Masking
    • SoftMask and BinaryMask

Tutorials

• Tutorials
  • Running REPET
    • Introduction
    • Repeating Period
    • Running Repet
    • Output of REPET
  • Evaluating performance
    • REPET
    • 2DFT
    • HPSS
    • Putting it all together
  • Handling data
    • BaseDataset
    • MUSDB18
    • MixSourceFolder
    • WHAM!
    • Scaper
    • Transforms
  • Training models
    • SeparationModel
      • Modules
      • Connections
      • Output and forward pass
      • Putting it all together
      • Saving and loading a model
    • Custom modules
    • Handling data
    • Closures and loss functions
      • Loss dictionary
      • Custom loss functions
      • Optimizing the model
    • Ignite Engines
      • Putting it all together
    • What’s in a model?
    • Using and evaluating a trained model
      • Evaluation in parallel

Examples

• Examples
  • Benchmark
    • HighPassFilter
    • IdealBinaryMask
    • IdealRatioMask
    • WienerFilter
    • MixAsEstimate
  • Deep learning
    • DeepClustering
    • DeepMaskEstimation
  • Composite methods
    • EnsembleClustering
  • Factorization
    • RPCA
    • ICA
  • Primitives
    • FT2D
    • REPET
    • REPETSIM
    • Melodia
    • TimbreClustering
    • HPSS
  • Spatial
    • SpatialClustering
    • PROJET
    • DUET

Recipes

• Recipes
  • WHAM!
    • Ideal binary mask
      • Imports
      • Setting up
      • Evaluation
    • Ideal ratio mask
      • Imports
      • Setting up
      • Evaluation

API Documentation

• API Documentation

Citing nussl

• Citing nussl

Contributing

• Contribution Guide
  • General
  • Documentation
  • Adding your own algorithm

Troubleshooting

For bug reports and issues with this code, please see the github issues page. Please review open issues before contacting the authors.

Changelog

• Changelog

Indices and tables

• Index

• Module Index

• Search Page

Footnotes

1

Officially stands for “Northwestern University Source Separation Library”, but in our hearts nussl stands for “Needs Unmixing? Source Separation Library!”