admeynard / jefas

Joint Estimation of Frequency, Amplitude and Spectrum

Contents

This repository contains all the MATLAB and Python files that are necessary to perform the following algorithms:

• JEFAS: Joint Estimation of Frequency, Amplitude and Spectrum
• JEFAS-BSS: JEFAS-Blind Source Separation
• JEFAS-S: JEFAS-Synthesis

The above-mentioned algorithms are fully available in MATLAB. The MATLAB folder provides all available code with additional explanations, several examples, and scripts that produce the figures presented in our articles. The Python folder contains (for now) only the routines to implement JEFAS and JEFAS-S, as well as some examples.

JEFAS

The theoretical background relative to the algorithm can be found in the following paper:
A. Meynard and B. Torrésani, "Spectral Analysis for Nonstationary Audio", IEEE/ACM Transactions on Audio, Speech and Language Processing, vol. 26, pp. 2371–2380, Dec. 2018, available online.

JEFAS-BSS

The theoretical background relative to the algorithm can be found in the following paper:
A. Meynard, "Spectral Estimation for Multivariate Locally Time-Warped Signals", IEEE Signal Processing Letters, vol. 27, pp. 311-315, 2020, available online.

JEFAS-S

The theoretical background relative to the algorithm can be found in the following paper:
A. Meynard and B. Torrésani, "Synthesis-based time-scale transforms for non-stationary signals ", submitted, 2022, available online.

Authors

Authors: Adrien Meynard and Bruno Torrésani
Contact email: [email protected]