基于DSP Builder的语音端点检测的实现

介绍了一种基于能量变迁的语音端点检测方法以及如何用DSP Builder实现该方法的硬件电路设计.设计时采用了查表、多时钟频率设计和有限状态机的方法提高运算速度,简化硬件结构.通过仿真和FPGA验证结果表明,该方法可以有效的检测出语音信号的起点和终点,动态分帧的结构可以满足嵌入式系统的实时性要求.

Implementation of Speech Endpoint Detection Based on DSP Builder

A method of speech endpoint detection base on energy state and how to design its hardware with DSP Builder are introduced in this paper.To obtain high speed and simplify the structure,many techniques are used,such as look-up table,multiply clock and FSM.The results of simulation and FPGA verification show that this design performs very efficiently on speech endpoint detection.In addition,the structure of dynamic enframing will meet the real-time requirement of embedded systems.