Blind Separation of Noisy Mixtures of Non-stationary Sources Using Spectral Decorrelation

Authors:
Hicham Saylani;Shahram Hosseini;Yannick Deville
Affiliations:
Laboratoire d'Astrophysique de Toulouse-Tarbes, Université de Toulouse, CNRS, Toulouse, France 31400 and Laboratoire des Systèmes de Télécommunications et Ingénierie de la ...;Laboratoire d'Astrophysique de Toulouse-Tarbes, Université de Toulouse, CNRS, Toulouse, France 31400;Laboratoire d'Astrophysique de Toulouse-Tarbes, Université de Toulouse, CNRS, Toulouse, France 31400
Venue:
ICA '09 Proceedings of the 8th International Conference on Independent Component Analysis and Signal Separation
Year:
2009

Citing 3
Cited 0

Second Order Nonstationary Source Separation

Journal of VLSI Signal Processing Systems
Blind separation of linear instantaneous mixtures of non-stationary signals in the frequency domain

Signal Processing
A blind source separation technique using second-order statistics

IEEE Transactions on Signal Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we propose a new approach for blind separation of noisy, over-determined, linear instantaneous mixtures of non-stationary sources. This approach is an extension of a new method based on spectral decorrelation that we have recently proposed. Contrary to classical second-order blind source separation (BSS) algorithms, our proposed approach only requires the non-stationary sources and the stationary noise signals to be instantaneously mutually uncorrelated. Thanks to this assumption, it works even if the noise signals are auto-correlated. The simulation results show the much better performance of our approach in comparison to some classical BSS algorithms.