用于车载无线终端的嵌入式语音处理系统  

Embedded Speech Processing System for Car Wireless Terminals

作　　者：刘志[1] 刘加[1] 刘润生[1] 

机构地区：[1]清华大学电子工程系,北京100084

出　　处：《计算机工程》2005年第6期182-183,202,共3页Computer Engineering

基　　金：国家自然科学基金资助项目(60272016)

摘　　要：介绍了一个用于汽车环境的无线终端中利用语音技术进行语音识别拨号、语音合成和语音提示的系统。系统包括两个主要的模块:语音处理模块和蓝牙通信模块。其中蓝牙通信模块的功能是与具有蓝牙接口的手机进行通信,包括连接手机进行通话,下载手机内的电话号码本并传送给语音处理模块;语音处理模块完成语音识别、语音合成、语音提示、利用语音压缩编解码进行通话录放以及号码查询等功能,并控制整个系统的流程。该系统可以实现对手机电话本的下载并在线生成识别词表,识别词表容量可达1000词;在600词情况下的实验结果表明系统的识别率大于97%;系统基于SoC架构,具有高集成度和高稳定性的特点。This paper introduces a novel embedded speech processing system, which is implemented in car wireless terminals. This system is composed of two modules: bluetooth module and speech-processing module. Through the blue tooth module, car wireless terminal can communicate with cellular phone, download the address book from cellular phone and send it to the speech processing module, while the speech processing module fulfills speech recognition, speech synthesis, speech coding and system controlling functions. This system can on-line generate the speech recognition network according to the address book and the recognizable vocabulary scale is up to 1000 words. When the address book includes 600 names, the recognition accuracy rate is more than 97%. The hardware platform of this system is based on SoC architecture and high integration of this architecture makes the system more robust.

关 键 词：语音处理 嵌入式系统 车载无线终端 SOC 

分 类 号：TN912.34[电子电信—通信与信息系统]