Blind extraction of underdetermined mixtures via time-frequency mask

Abstract

For blind source estimation problem of underdetermined and non-completely sparse signals,a novel blind extraction algorithm was proposed.The algorithm combined the non-completely sparse blind extraction algorithm with the DUET approach.It computed firstly the extraction vector of source and the normal vector of source direction,and then renewed the mixtures and the mixing matrix using a linear transform,and recoverd the source using time-frequency mask method finally.Several speech signal experiments demonstrate its performance and practical.

Keywords

• underdetermined blind source separation
• sparse component analysis
• time-frequency mask
• blind source extraction