Voice activity detector for microphone array processing in hand-free systems

This paper describes the use of voice activity detector (VAD) in microphone array processing for hand-free systems. Two different types of VAD are proposed using conventional and adaptive beamforming. Taking into account the fact that the speech signal is not present at all time in human conversations, both signal of interest and noise can be better estimated. Simulation results show that the VAD significantly improves the performance of the array processing algorithms. The performances of the two types of VAD are also compared in the paper.