A new speech enhancement method

The paper proposes a speech enhancement method in the wavelet domain, that is, using a soft thresholding to remove noise components from the wavelet coefficients of noisy speech. To prevent the quality degradation of the unvoiced sounds during the denoising process, we first separate unvoiced region from noisy speech and then apply the thresholding method in a different way from other regions. Simulation results demonstrate that the method has good performance. Considering the simplicity of the proposed algorithm, this method is proven to be very promising.