基于FPGA的嵌入式语音识别控制系统

介绍了一款针对特定人、孤立词的嵌入式语音识别系统的设计与实现.该系统的硬件核心部件是Virtex-II Pro50 FPGA芯片,其硬核处理器是PowerPC405.本系统对预处理、端点检测、LPCC特征提取部分进行了定点化处理;DTW算法采用硬件IP核实现;整体调度采用中断方式实现.将该系统用于语音控制玩具机器狗AIBO进行实验,识别率达到98.3%.本系统设计性能满足玩具、游戏等消费娱乐电子设备对识别率和实时性的性能要求,具有广阔的市场应用前景.

Embedded Speech Recognition Control System Based on Field Programmable Gate Array

This paper introduces the design and implementation of a speaker dependent, isolated word embedded speech recognition system. The core part of its hardware is Virtex-II Pro50 FPGA chip, with PowerPC405 as its hard processor core. The system uses Fixed-Point Processing to implement Pretreatment, Endpoint Detection and LPCC Feature Extraction. IP core is used to implement DTW algorithm. Interrupt is used to implement main program dispatch. The speech recognition accuracy reaches 98.3% when using the system to do test to control the AIBO Dog toy. The performance of this system can meet the accuracy and time requirements of speech recognition system, so it can be widely used to control such entertainment electronic equipment as toys, games, and so on. It has great potential in market application.