AUDIO SOURCE SEPARATION USING SPARSITY

Abdeldjalil Aissa El Bey (1) , Hicham Bousbia-Salah (2) , Karim Abed-Meraim (1) , Yves Grenier (1)

1 ENST - Ecole Nationale Supérieure des Télécommunications
2 ENP - École nationale polytechnique [Alger, Algérie]

In this paper, we are interested in blind source separation from instantaneous mixtures of audio signals. Using the sparsity property of audio signals, we propose an iterative method that relies on a relative gradient technique which minimizes a contrast function based on the p norm. This norm is considered as a good sparsity measure. The simulations show that the proposed method outperforms other methods based on source independency.

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

Abdeldjalil Aissa El Bey, Hicham Bousbia-Salah, Karim Abed-Meraim, Yves Grenier. AUDIO SOURCE SEPARATION USING SPARSITY. 10th IEEE International Workshop on Acoustic Echo and Noise Control (IWAENC), Sep 2006, Paris, France. ⟨hal-01772652⟩