UNDERDETERMINED BLIND SEPARATION OF AUDIO SOURCES IN TIME-FREQUENCY DOMAIN

Abdeldjalil Aissa El Bey (1) , Karim Abed-Meraim (1) , Yves Grenier (1)

1 ENST - Ecole Nationale Supérieure des Télécommunications

This paper considers the blind separation of audio sources in the underdetermined case, where we have more sources than sensors. A recent algorithm applies time-frequency distributions (TFDs) to this problem and gives good separation performance in the case where sources are disjoint in the time-frequency (TF) plane. However, in the non-disjoint case, the reconstruction of the signals requires some interpolation at the intersection points in the TF plane. In this paper, we propose a new algorithm that combines the abovementioned method with subspace projection in order to explicitly treat non-disjoint sources. Another contribution of this paper is the estimation of the mixing matrix in the underdetermined case.

Domaines: Traitement du signal et de l'image [eess.SP]

Abdeldjalil Aissa El Bey, Karim Abed-Meraim, Yves Grenier. UNDERDETERMINED BLIND SEPARATION OF AUDIO SOURCES IN TIME-FREQUENCY DOMAIN. Signal Processing with Adaptative Sparse Structured Representations (SPARS), Nov 2005, Rennes, France. ⟨hal-01771658⟩