Noise reduction method based on generalized subtractive beamformer

Abstract

Combating noise signals is still a very important and challenging research topic. Previously, we presented a subtractive-beamformer-based noise reduction algorithm using paired microphones, which was shown to be effective in reducing directional noise. However, its basic assumption, namely a perfectly coherent noise field, is generally not satisfied in real-world environments. In this paper, we develop a general expression for the original algorithm we suggested earlier, on the basis of a generalized subtractive beamformer and by relaxing the strict assumption to that of an arbitrary noise field. Following ideas similar to those of the original algorithm, the generalized algorithm with a generalized sidelobe canceller (GSC)-like structure is derived. A theoretical analysis is then presented to show the linkage between this generalized algorithm and the original algorithm, and to show its noise reduction performance in theoretically defined noise fields. Finally, the superiority of the proposed noise reduction algorithm to other comparable algorithms was verified by experiments using multichannel recordings.