Speech Endpoint Detection Algorithm for Uyghur Based on Acoustic Frequency Feature

Accurate endpoint detection is important for improving the speech recognition capability. This paper proposes an effective endpoint detection algorithm based on the acoustic frequency feature for Uyghur. The spectrum of each speech frame is divided into several sub-bands, and the maximum average spectral density of these sub-bands is used as the detection criteria to distinguish the speech and noise. At the same time, a dynamically updated threshold and a smoothing window are used to improve the performance of the algorithm. It is characterized by higher accuracy or flexibility, faster processing speed and less computation. Experimental results show that the proposed algorithm achieves better performance compared to energy based and zero-crossing rate based algorithms.