Speech acquired from an array of distant microphones is affected by ambient noise and reverberation. Single channel linearly constrained minimum variance (LCMV) filters have been proposed to remove ambient noise. In this paper, an algorithm for joint noise cancellation and dereverberation using a multi channel LCMV filter in the frequency domain is proposed. A single channel LCMV filter which accounts for the inter frame correlation is applied on each channel to remove the early reverberation component. A modified spectral subtraction method is also proposed to remove the late reverberation component present in the speech signal. Experimental results on joint noise cancellation and dereverberation indicate a reasonable improvement over conventional speech enhancement methods. Additional experiments on distant speech recognition are also conducted to illustrate the significance of the method.
Bibliographic reference. Nathwani, Karan / Hegde, Rajesh M. (2013): "Joint noise cancellation and dereverberation using multi-channel linearly constrained minimum variance filter", In INTERSPEECH-2013, 2987-2991.