Frequency Domain Blind Source Separation for Many Speech Signals

This paper presents a method for solving the permutation problem of frequency domain blind source separation (BSS) when the number of source signals is large, and the potential source locations are omnidirectional. We propose a combination of small and large spacing sensor pairs with various axis directions in order to obtain proper geometric information for solving the permutation problem. Experimental results in a room (reverberation time TR=130 ms) with eight microphones show that the proposed method can separate a mixture of six speech signals that come from various directions, even when two of them come from the same direction.

• Type: Conference
• Year: 2004
• Where: ICA
• Authors: Ryo Mukai, Hiroshi Sawada, Shoko Araki, Shoji Makino
• Keywords: Blind Source Separation, ICA 2004, Large Spacing Sensor, Permutation Problem