Multizone near-end speech enhancement under optimal second-order magnitude distortion

In this article, we address near-end speech enhancement for a scenario where there are several playback zones. A signal model is explored, where effects of noise, reverberation and zone crosstalk are taken into account simultaneously. Through the symbolic usage of a general smooth distortion measure, necessary optimality conditions are derived. The conditions are applied to a DFT magnitude-based distortion measure and an algorithm follows, which applies per-zone spectral subtraction followed by channel inversion. Simulations validate the optimality of the algorithm and show a clear benefit in multizone processing, as opposed to the iterated application of a single-zone algorithm.