Short-time multichannel noise correlation matrix estimators for acoustic signals

The estimation of noise power spectral densities is a critical step in many speech enhancement algorithms. The demand for multi-channel speech enhancement systems is increasing with applications such as teleconferencing, cellular phones and hearing aids. The objective of this paper is to present a general multi-channel framework to solve for the diffuse noise correlation matrix whenever the spatial spectral matrix or coherence matrix is pre-estimated and the number of active speakers is less than the number of microphones. The performance of the developed algorithms is compared with pre-existing algorithms using prescribed performance measures.