University of Ottawa, December 6 - 9, 2013
In this talk, I will review the algorithm DUET that addresses the blind
source separation problem. In addition to being computationally efficient, one of the advantages of DUET is its ability, at least in some cases, to separate $n\ge 3$ source signals using only two mixtures. The algorithm is based on the key
observation that Gabor expansions of speech signals are approximately
sparse and satisfy the so-called $W$-disjoint orthogonality assumption which states that which states that, statistically, the windowed Fourier transforms of different source signals have disjoint supports.