基于DSP的高速实时语音识别系统的设计与实现  

Realization and Design of High Speed Real-time Speech Recognition Based on DSP

作　　者：李邵梅[1] 陈鸿昶[1] 王凯[1] 

机构地区：[1]国家数字交换系统工程技术研究中心,河南郑州450002

出　　处：《现代电子技术》2007年第15期109-111,共3页Modern Electronics Technique

基　　金：国家自然科学基金(60372038)

摘　　要：识别正确率和抗噪性能固然是语音识别的研究重点,但是识别响应速度也是决定系统实用化的关键所在。以TMS320C6713为核心构建硬件平台,通过采用高效C语言和线性汇编混合编程的方式,结合硬件特点,对代码进行了优化,实现了以美尔频率倒谱系数为特征参数,采用动态时间弯折算法的高速语音实时识别系统,识别速率达0.29倍实时,可实现多路语音的并行识别。Response rate of recognition is a key factor for a speech recognition system as well as recognition correct rate and noise robust property. An embedded speech recognition system is designed and realized in this paper. Its hardware＇s core processor is TMSDC6713. Speech recognition system＇s software consists of several modules such as computation of MFCC,dynamic time warping algorithm,judgment of recognition result, etc. This paper introduces some effective ways to optimize C language,including the hardware feature, elimination of dependency among instruction registers, algorithm optimization and improvement of code structure, and gives the experiment result.

关 键 词：多路实时语音识别 数字信号处理器 美尔频率倒谱系数 动态时间弯折 

分 类 号：TP29[自动化与计算机技术—检测技术与自动化装置]