Speech enhancement based on the generalized sidelobe cancellation and spectral subtraction for a microphone array

Abstract

GSC (Generalized sidelobe canceller) algorithm is a widely used speech enhancement method for a microphone array. Whereas, The GSC has some problems such as: speech offset in adaptive noise cancellation module is weak, noise cancellation capability for non-coherent noise is not enough. Considering the GSC's problems of the noise cancellation capability and the residual background noise, the paper proposes a new combination of GSC and spectral subtraction speech enhancement method. The simulation results show that this method can effectively suppress the influence of the residual background noise and improve speech-noise ratio and intelligibility of speech.