Analysis of two-sensors forward BSS structure with post-filters in the presence of coherent and incoherent noise

We consider the speech enhancement problem in a moving car through a blind source separation BSS scheme involving two sensors. To correct the distortion brought by this structure we have proposed in previous work (Djendi et al., 2007) two frequency-domain methods to compute the post-filters placed at the output of the forward BSS structure (FBSS). In this work, we consider the case where the noises at the sensor inputs contain coherent and non-coherent components. We provide an analysis of the performance (output SNR and the distortion criterion) of the FBSS structure with post-filters as a function of two new parameters: the coherent to diffuse ratio (CDR) and the speech to coherent ratio (SCR). Simulation results show perfect agreement between theoretical and experimental results. Crown Copyright (c) 2013 Published by Elsevier B.V. All rights reserved.