A Blind Algorithm for Joint Noise Suppression and Dereverberation

Abstract:
This contribution presents a novel speech enhancement algorithm for the suppression of background noise and late reverberation without a priori knowledge. The speech enhancement is performed by a generalized spectral weighting rule based on spectral power estimations for the late reverberant speech and background noise. By this, speech distortions due to late room reflections can be suppressed without knowledge about the complete room impulse response. In contrast to existing methods, all needed quantities are estimated entire blindly from the reverberant and noisy speech signal. A possible application of the proposed algorithm are speech enhancement systems for hands-free devices, mobile phones or hearing aids.