FastICA

Technologies

Project was created and tested with:

Windows 10
Python 3.6.5

Description

Project used to split N mixed audio files into N separated independent audio files. This project uses FastICA algorthm (not the most accurate one). Currently this project supports only .wav audio files.

Setup

Run following block of commands in fastICA\ catalogue:

python -m virtualenv venv
cd venv
cd Scripts
activate
cd ..
cd ..
pip install -r requirements.txt

Set all paths for audio files in "Editable parameters" section in fastica.py script
Set all parameters in "Editable parameters" section in fastica.py script

Run

Go to fastICA\ and run command:

python fastica.py

References

This project is created based on following documents:

https://eti.pg.edu.pl/documents/176593/26756916/STS.pdf (pages 43-44)
A. Hyv¨arinen, E. Oja. “A fast fixed-point algorithm for independent component analysis”, Neural Computation, vol. 9, pp. 1483-1492, 1997.