Blind convolutive separation algorithm for speech signals via joint block diagonalization

A blind speech source separation algorithm for the overdetermined convolutive mixture model in time-domain is proposed via joint block-diagonalization based on the mutual-independence property and the short-time stationary of the speech signals.Taking the sum of the F -norms of all off-diagonal sub-matrices as a criterion,a novel joint block-diagonalization algorithm is proposed to estimate the whole mixture matrix through minimizing a sequence of quadratic subfunctions corresponding to mixture submatrices.Both theoretical analysis and simulation results show that the proposed algorithm has much lower complexity and faster convergence speed than the classical Jacobi-like method with no performance loss.In addition,there almost are no obvious impacts of the channel order and initialization values on the convergence speed.