基于临界带功率谱方差的端点检测

端点检测作为语音信号处理的关键技术，其准确性直接影响到语音识别系统的计算复杂度和识别能力。在人耳听觉特性理论研究的基础上，利用语音段和背景噪声段临界带功率谱上的差异，提出了一种基于临界带功率谱方差的端点检测方法。通过自适应门限值的选取，该方法对背景噪声具有良好的跟踪性能。在不同的信噪比条件下，进行了端点检测实验。结果表明：该方法与传统的短时能量和短时平均过零率方法、谱熵方法相比，可以有效降低背景噪声的影响，具有更好的鲁棒性和正确率。

A voice activity detection algorithm based on the variance of critical band power spectrum

As the key technology of speech signal processing,the accuracy of voice activity detection directly affects the computational complexity and identification capabilities of speech recognition system.Based on theoretical research of human auditory characteristics,this paper presents a voice activity detection algorithm which involves the variance of critical band power spectrum.By using adaptive threshold,the presented method has a good performance on the tracking of background noise.Experiments are taken at different signal to noise ratios.The results show that comparing with short time features-based algorithm and entropy based algorithm,this method can effectively reduce the influences of background noise.Furthermore,it also has better robustness and higher accuracy.