【摘要】 设计一种以TM S320VC5402为核心,外围扩展语音编解码器、程序存储器、语音存储器等集成电路,基于定点数字信号处理器的实时说话人身份识别的系统.通过为每个语音命令设置3个模板的预留空间,使每个语音命令可以有3个模板参与识别,避免说话人语气和语调的变化产生的影响.在比较嘈杂的环境下,系统的正确识别率超过87%.处理器VC5402从识别中断并响应中断到CPU进入中断服务程序的第1条指令,需要20个时钟周期,系统的平均处理时间小于0.2 s.更多还原

【Abstract】 Designing a real-time speaker recognition system that is based on the fixed-point digital signal processor.The system is adopted the TMS320VC5402 as the handle center,its peripheries are speech codec,program memory,speech memory,etc.integrated circuits.By means of setting 3 templates reserved space for each speech command,making every speech command may have 3 templates participate identification,which can avoid the influence of speaker’s tone and intonation changing.In fairly noise conditions,the system can have correct recognition over 87%.The processor of VC5402 needs 20 clock periods from recognize and respond interrupt to CPU enter interrupt serve program,the average time of system processing is less than 0.2 second.更多还原