Subband dereverberation algorithm for noisy environments

In hands-free voice applications (including voice communication and voice recognition), the captured signal is severely degraded by reverberation when a talker is further away from the microphone. Reverberation significantly degrades speech quality and speech recognition accuracy. As a result, it is necessary to suppress reverberation. Spectral subtraction based approach has been used for reverberation reduction. In this paper, an efficient subband de-reverberation approach based on spectral subtraction is proposed. The Algorithm is able to effectively reduce reverberation in noisy environments while maintaining low complexity and low signal distortion level.