一种基于短时平均幅度差的语音检测算法

提出了一种利用语音信号的短时平均幅度差特征并结合短时平均幅度的语音检测算法。该算法在专用通信系统中,用以对接收到的电台信号进行分析,判断其中是否有语音信号,从而控制半双工电台的发射开关,使其处于接收或发射状态。实验表明,该算法能在较低的信噪比情况下准确地检测出语音信号,而且计算方法简单,硬件处理容易,可靠性高,能够满足实时系统的需要。

Research and Realization of Voice Active Detection Based on Short-time AMDF

A new simple and reliable voice active detection algorithm which utilizes the characteristics of short-time AMDF(Average Magnitude Difference Function) and short-time AM(Average Magnitude) to detect the speech endpoint is presented. The algorithm is used in specially designed communication system which analyzes the received radio station signals, estimates whether there are voice signals and controls the sending switch of the semi duplex radio station which would be changed between receiving and sending states. It is proved that this algorithm has high performance even in a low signal-to-noise ratio circumstance and has the advantages of simple computing and easy hardware processing. Its high reliability makes it suitable for real-time communication.