The estimation of fundamental frequency of speech using microphone array

In this paper we propose a simple and robust method for estimating the fundamental frequency of speech when multiple speech signals and background noise exist simultaneously. The method can be realized by using delay-sum beamforming and blocking processing, such as blocking matrix operation of the generalized sidelobe canceller (GSC), and it is based on the difference of the output harmonic components of those two processing. The proposed method is shown to estimate the fundamental frequency of the desired speech effectively under the severe environmental conditions. After the estimation, the delay-sum beamformer output is processed with a comb filter to extract the desired speech.