基于HMM模型的非特定人车载音响语音控制系统

The Speaker-Independent Speech Recognition System for Audio System of Automotive Using HMM Model

提出一种语音命令控制车载音响操作的设计方案，以德国Infineon公司新推出的具有DSP和单片机双核的Soc语音处理芯片UniSpeech—SDA80D51为核心组成非特定人车载音响语音控制器系统，实现对SL1102C1车载音响的语音控制。介绍了语音控制器系统的构成、专用语音处理芯片SDA80D51功能及工作原理、非特定人语音识别软件HMM算法以及SDA80D51与SL1102C1接口控制方案。系统在江淮同悦SL1102C1型车载音响上进行了语音控制实测，实测数据表明系统语音识别率可达到95％。

This paper has propose a design of speech control for audio system of automotive. The Infineon＇s new Soc speech processing chip UniSpeech-SDASOD51 with two centers, DSP and MCU, which as the center of automotive-audio speech control system, realize the speaker-independent speech recognition and control of automotive-audio SL1102C1. Introduced the structure of speech control system, SDASODS1, the Speaker-lndependent speech recognition system using HMM Mode and the interface of SDASODS1 with SL1102C1. realize the speaker-independent speech recognition and control of automotive-audio SLll02C1 on TonYue JAC. After testing the system in the relatively quiet environment, speech recognition accuracy rate can reach 95%.