A new speech enhancement method for adverse noise environment

Abstract

A new speech enhancement method combined independent component analysis (ICA) with delay-sum beamforming is presented in this paper. The noise signals are separated from the speech signal by using ICA module whose inputs are the microphone array signals after time delay compensation. Then the output of the delay-sum beamforming and the separated noise signals are executed adaptive noise canceling according to the criterion of minimum output energy (MOE). Through this method, relatively 'pure' speech signal can be obtained in directional noise field or uncorrelated noise field. Some simulation results in the presence of different signal-to-noise ratio (SNR) are shown to demonstrate the validity of the proposed method especially in adverse noise environment. © Springer-Verlag Berlin Heidelberg 2005.