Implementation effective one-channel noise reduction system

This contribution addresses the problem of additive noise reduction using one-channel noise suppression system. A new implementation effective method is suggested and evaluated. The method consists of two independent parts. The noise estimation part is based on the noise matched filter producing an estimation of background noise without the need of a voice activity detector. The noise reduction part uses short-time spectral attenuation technique. The main idea reducing computational costs lies in the use of reduced number of frequency bands for computation of attenuation factors. Except of reduced number of operations this approach decreases fluctuations of estimated spectral gains, and therefore the speech distortion is low. Thus the suggested system eliminates the need of any enhanced speech postprocessing. A new effective approach is used for the inverse frequency transformation. In spite of the simplicity of the suggested method its performance is comparable with other existing one-channel noise reduction methods.