Two time-frequency ratio-based blind source separation methods for time-delayed mixtures

Matthieu Puigt (1) , Y. Deville (1)

1 LATT - Laboratoire Astrophysique de Toulouse-Tarbes

We propose two time-frequency (TF) blind source separation (BSS) methods suited to attenuated and delayed (AD) mixtures. They consist in identifying the columns of the (filtered permuted) mixing matrix in Constant-Time TF zones where they detect that a single source occurs, using TIme-Frequency Ratios Of Mixtures (hence their name AD-TIFROM-CT). We thus identify columns of scale coefficients and time shifts. Unlike various previously reported TF-BSS approaches, these methods set very limited constraints on the source sparsity and overlap. They are especially suited to non-stationary sources.

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

Matthieu Puigt, Y. Deville. Two time-frequency ratio-based blind source separation methods for time-delayed mixtures. Proceedings of the Sixth International Conference on Independent Component Analysis and Blind Signal Separation (ICA 2006), Mar 2006, Charleston, United States. pp.682-690, ⟨10.1007/11679363⟩. ⟨hal-00289252⟩