Ray-Space Constrained Multichannel Nonnegative Matrix Factorization for Audio Source Separation

This repository contains the implementation of the Ray-Space Constrained Multichannel Nonnegative Matrix Factorization (MNMF) method for audio source separation. This approach extends conventional MNMF by integrating a Ray-Space model, enabling a more precise representation of spatial characteristics and improving source separation performance.

Features

• Ray-Space Dictionary Construction: Models the propagation of signals using Green’s functions and applies a Ray-Space transform.
• Frequency-Dependent Propagation Modeling: Addresses the limitations of traditional far-field models by incorporating frequency dependency.
• Regularized Source Activation: Introduces constraints to prevent multiple sources from being assigned to the same grid position simultaneously.

Installation

Clone this repository and set up the required dependencies:

git clone https://github.com/your_username/ray-space-constrained-mnmf.git
cd ray-space-constrained-mnmf

Ensure that MATLAB is installed on your system and add the repository path to MATLAB using the following span

Usage

Run the main.m script to perform source separation on an example dataset

Contributing

Contributions are welcome! To contribute:

1. Fork the repository.
2. Create a feature branch (git checkout -b feature-name).
3. Commit your changes (git commit -m "Add feature").
4. Push to your branch (git push origin feature-name).
5. Open a pull request.

Reference

If you use this implementation in your research, please cite:

A. J. Muñoz-Montoro, M. Olivieri, M. Pezzoli, J. Carabias-Orti, F. Antonacci and A. Sarti, "Ray-Space Constrained Multichannel Nonnegative Matrix Factorization for Audio Source Separation," 2024 32nd European Signal Processing Conference (EUSIPCO) , Lyon, France, 2024, pp. 396-400, doi: 10.23919/EUSIPCO63174.2024.10715403.

License

This project is licensed under the MIT License.