Skip to content

tky823/ssspy

Repository files navigation

ssspy

Documentation Status tests codecov Open in Spaces

A Python toolkit for sound source separation.

Installation

You can install by pip.

pip install ssspy

To install latest version,

pip install git+https://github.com/tky823/ssspy.git

Instead, you can build package from source.

git clone https://github.com/tky823/ssspy.git
cd ssspy
pip install .

If you cannot install ssspy due to failure in building wheel for numpy, please install numpy in advance.

Build Documentation Locally (optional)

To build the documentation locally, you have to include docs and notebooks when installing ssspy.

pip install -e ".[docs,notebooks]"

You need to convert some notebooks by the following command:

# in ssspy/
. ./docs/pre_build.sh

When you build the documentation, run the following command.

cd docs/
make html

Or, you can build the documentation automatically using sphinx-autobuild.

# in ssspy/
sphinx-autobuild docs docs/_build/html

Blind Source Separation Methods

Method Notebooks
Independent Component Analysis (ICA) [1-3] Gradient-descent-based ICA: Open in Colab
Natural-gradient-descent-based ICA: Open in Colab
Fast ICA: Open in Colab
Frequency-Domain Independent Component Analysis (FDICA) [4-6] Gradient-descent-based FDICA: Open in Colab
Natural-gradient-descent-based FDICA: Open in Colab
Auxiliary-function-based FDICA (IP1): Open in Colab
Auxiliary-function-based FDICA (IP2): Open in Colab
Gradient-descent-based Laplace-FDICA: Open in Colab
Natural-gradient-descent-based Laplace-FDICA: Open in Colab
Auxiliary-function-based Laplace-FDICA (IP1): Open in Colab
Auxiliary-function-based Laplace-FDICA (IP2): Open in Colab
Independent Vector Analysis (IVA) [7-14] Gradient-descent-based IVA: Open in Colab
Natural-gradient-descent-based IVA: Open in Colab
Fast IVA: Open in Colab
Faster IVA: Open in Colab
Auxiliary-function-based IVA (IP1): Open in Colab
Auxiliary-function-based IVA (IP2): Open in Colab
Auxiliary-function-based IVA (ISS1): Open in Colab
Auxiliary-function-based IVA (ISS2): Open in Colab
Auxiliary-function-based IVA (IPA): Open in Colab
Gradient-descent-based Laplace-IVA: Open in Colab
Natural-gradient-descent-based Laplace-IVA: Open in Colab
Auxiliary-function-based Laplace-IVA (IP1): Open in Colab
Auxiliary-function-based Laplace-IVA (IP2): Open in Colab
Auxiliary-function-based Laplace-IVA (ISS1): Open in Colab
Auxiliary-function-based Laplace-IVA (ISS2): Open in Colab
Auxiliary-function-based Laplace-IVA (IPA): Open in Colab
Gradient-descent-based Gauss-IVA: Open in Colab
Natural-gradient-descent-based Gauss-IVA: Open in Colab
Auxiliary-function-based Gauss-IVA (IP1): Open in Colab
Auxiliary-function-based Gauss-IVA (IP2): Open in Colab
Auxiliary-function-based Gauss-IVA (ISS1): Open in Colab
Auxiliary-function-based Gauss-IVA (ISS2): Open in Colab
Auxiliary-function-based Gauss-IVA (IPA): Open in Colab
Independent Low-Rank Matrix Analysis (ILRMA) [15-18] Gauss-ILRMA (IP1/MM): Open in Colab
Gauss-ILRMA (IP1/ME): Open in Colab
Gauss-ILRMA (IP2/MM): Open in Colab
Gauss-ILRMA (IP2/ME): Open in Colab
Gauss-ILRMA (ISS1/MM): Open in Colab
Gauss-ILRMA (ISS1/ME): Open in Colab
Gauss-ILRMA (ISS2/MM): Open in Colab
Gauss-ILRMA (ISS2/ME): Open in Colab
Gauss-ILRMA (IPA/MM): Open in Colab
Gauss-ILRMA (IPA/ME): Open in Colab
t-ILRMA (IP1/MM): Open in Colab
t-ILRMA (IP1/ME): Open in Colab
t-ILRMA (IP2/MM): Open in Colab
t-ILRMA (IP2/ME): Open in Colab
t-ILRMA (ISS1/MM): Open in Colab
t-ILRMA (ISS1/ME): Open in Colab
t-ILRMA (ISS2/MM): Open in Colab
t-ILRMA (ISS2/ME): Open in Colab
GGD-ILRMA (IP1/MM): Open in Colab
GGD-ILRMA (IP2/MM): Open in Colab
GGD-ILRMA (ISS1/MM): Open in Colab
GGD-ILRMA (ISS2/MM): Open in Colab
Independent Positive Semidefinite Tensor Analysis (IPSDTA) [19, 20] Gauss-IPSDTA (VCD): Open in Colab
t-IPSDTA (VCD): Open in Colab
Multichannel Nonnegative Matrix Factorization (MNMF) [21-24] Gauss-MNMF: Open in Colab
t-MNMF: soon
Fast Gauss-MNMF (IP1): Open in Colab
Fast Gauss-MNMF (IP2): Open in Colab
Blind Source Separation via Primal-Dual Splitting Algorithm (PDS-BSS) [25,26] PDS-BSS: Open in Colab
PDS-BSS-multiPenalty: Open in Colab
PDS-BSS-masking: Open in Colab
Blind Source Separation via Alternating Direction Method of Multipliers (ADMM-BSS) ADMM-BSS: Open in Colab
Harmonic Vector Analysis (HVA) [27] HVA: Open in Colab
Complex Angular Central Gaussian Mixture Model (cACGMM) [28] cACGMM: Open in Colab

LICENSE

Apache License 2.0