Use of coherence in wavelet and multiwavelet domain for microphone array speech enhancement

In this paper, we are interested in multichannel speech denoising in the context of mobile communications. The conventional method exploits the "similarity" between the observations expressed by the coherence function. In this work, we aim at alleviating the drawbacks of this approach. More precisely, a wavelet or a multiwavelet transform is used to generate a multiresolution representation of the observations and the coherence between the resulting coefficients is computed. Experiments carried out on artificially and naturally noisy signals indicate that significant gains are achieved by the proposed method relatively to the classical coherence denoising technique.