基于SG/Simulink语音端点检测的设计与实现

Design and implementation of speech endpoint detection based on SG/Simulink

为了提高语音端点检测硬件电路的实时性,提出了一种基于短时幅度变化的语音端点检测方法。根据该检测方法,使用赛灵思公司的数字处理系统开发工具实现了端点检测硬件电路的设计。设计中采用了查找表和有限状态机等方法弥补了数字处理系统开发工具中元件库不足的缺点,从而简化了硬件设计,提高了设计的速度。通过协同仿真和现场可编程门阵列元件下载验证表明,该方法可以有效地检测出语音信号的起止点,硬件电路的设计满足系统有效性和准确性的要求。

To improve the real-time requirement of the endpoint detection hardware circuit,a method of speech endpoint detection based on short-time amplitude state is presented.According to the detection method,the Xilinxs’ development tools of digital processing system（system generator,SG） are used to achieve the hardware design.In the design,many methods are used,such as look-up table,FSM,and to make up for the shortage of the component database.And the development period is also shorten.Finally,by the results of co-simulation and FPGA verification,the validity of the presented design is demonstrated.The veracity and the efficiency requirement of the system are also met.