低端DSP芯片的语音处理算法研究

Research on Voice Processing Algorithm of Low-end DSP Chip

介绍了一种利用TI公司的TMS320VC5402定点DSP芯片实现的基于语音能量、短时平均幅度差和过零率的语音检测器,给出了算法的详细设计过程和DSP硬件实现方案,该方案在专用通信系统中,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,在对DSP在其他方面的应用也有一定的参考价值。

This paper proposes a voice detector implemented by using fixed-point DSP chip TMS320VC5402 from TI. Inc based on voice energy, short-time AMDF and zero crossing rate. The design details of algorithm and DSP hardware implement scheme are presented. The algorithm is used in special communication system to implement analysis on the received radio station signals and estimate whether there are voice signals, thus make the transmit switch of semiduplex radio station change between receive and transmit states. The experiment results show that this algorithm has high performance even in a low signal-to-noise ratio environment and such advantages as simple computing method, easy hardware processing and high reliability which can meet the requirements of real-time communication. It has some reference value to the other applications of DSP.