A new blind signal separation algorithm based on second-order statistics
Dwe Daniël Schobben, Pcw Piet Sommen · TU/e Research Portal · 1998
This paper addresses the problem of separating multiple speakers from mixtures of these that are obtained using multiple microphones in a room. A new adaptive blind signal separation algorithm is derived which is entirely based on Second Order Statistics which is entitled CoBliSS. The CoBliSS algorithm can run in offline or online (adaptive) mode. One of the advantages of the CoBliSS algorithm is that no assumptions are made about the probability density functions or other properties of the signals. Experiments with real recordings were carried out in a normal living room which show that the algorithm has good performance. As opposed to most other algorithms no parameters need to be tuned.