搜索
中国矿业大学大学生实践创新训练计划基于嵌入式应用的数字语音识别技术的研究项目级别：校级指导教师：殷春浩项目主持人：张浩、吴松、印魏、江磊、汪亚琦目录1.引言....................................................................................................22.设计要求..............................................................................................43.方案概述..............................................................................................54.模组特性简介...........................................................................……64.1SPCE061A特性简介..................................................................74.2芯片特性简介.............................................................................84.3SPCE061A精简开发板..............................................................94.451单片机驱动彩屏模组简介......................................................94.4.1彩屏驱动电路图......................................................................104.4.2主要功能..................................................................................104.4.3计算器实物图..........................................................................114.4.4注意事项..................................................................................115系统的硬件设计................................................................................125.1SPCE061A周边模块......................................................................125.1.1SPCE061A最小系统...................................................................135.1.2电源模块......................................................................................135.1.3放音模块......................................................................................145.1.4彩屏驱动电路..............................................................................146所用语音算法介绍............................................................................156.1语音识别算法简介.........................................................................156.1.1语音识别概述：........................................................................156.1.2语音识别原理..............................................................................156.1.3SPCE061A实现语音识别的步骤:..............................................156.1.4语音识别API介绍.....................................................................156.2凌阳音频概述..................................................................................166.2.1凌阳音频压缩算法的编码标准...................................................166.2.2压缩分类.......................................................................................166.2.3凌阳常用的音频形式和压缩算法..............................................166.2.4凌阳语音的播放、录制、合成和辨识......................................176.2.5常用的应用程序接口API的功能介绍及应用.........................177系统软件设计.....................................................................................247.1擦除FLASH模块：....................................................................247.2位操作模块：..................................................................................257.3语音播放函数..................................................................................277.4带权位数值结果播报函数模块......................................................28851单片机驱动彩屏显示模块..........................................................289应用方案的验证步骤.......................................................................2910结语...................................................................................................3011经费使用情况.....................................................................................3212参考文献.............................................................................................3213附录源代码.........................................................................................33【摘要】在现代社会语音识别这个词语已经不再新鲜，基于语音识别的计算器可让我们的计算省去手工输入麻烦。以SPCE061A单片机控制的语音识别模块，由于SPCE061A同时可分辨出最多命令为5条，作为语音识别计算器需要识别0~9加减乘除等命令远远超过5条，故采用分组命令形式，将0、1、2、3作为第一组，4、5、6、7作为第二组，8、9、清除、等于作为第三组，加、减、乘、除作为第四组，同时每组加上“下一组”命令用于切换语音识别当前所在的组。初始化时训练特定人的语音，将其语音特征值存储至SPCE061A的ROM，训练完毕后进入识别模式根据当前组的不同将，检测到语音输入通过对比特征值获取识别结果，并将识别结果用串口传送至51单片机彩屏显示模块，如果识别到“下一组”时，重新将下一组ROM的5个语音特征值装载到当前内存继续识别。识别到“等于”命令时计算出结果，并依照数值权位准确播报当前结果。【关键词】SPCE061A;凌阳单片机;特定人语音识别;语音识别计算器TheDesignoftheSpeechRecognitionCalculator【Abstract】Thespeechrecognitionisdividedintotwostages,namely,trainingandrecognition.Atthefirststage,thespeechrecognitionsystemlearnsaboutthehumanlanguageandstoreswhatithaslearnedinaspeechdatabase.Thenatthenextstage,themeaningofeachinputtedspeechcanimmediatelybefoundinthespeechdatabase.Sunplus16-bitSPCE061ASCMisembeddedwith32KwordFlashand2KwordSRAM,withbuilt-in10-bitADCandDACaswellasmorethan14interruptsources.ThecoreofitsCPUisa16-bitmicroprocessorchipwhichhasthefunctionofDSP.Besides,theCPUcanworkwithafrequencyupto49MHz,andprocesscomplexdigitalsignalseasilyandquickly.Therefore,comparedwithothertypesofSCM,SPCE061Ahasmoreadvantagesinspeechprocessing.BasedonSPCE061A,aspeechrecognitionCalculatorhasbeendesigned.Aftertraining,theCalculatorcancompletemanyCalculations,whichmakesthehuman-computerinteractionmoreintelligent.【Keywords】SPCE061ASCMSpeechRecognitionCalculator1引言在农业、金融业、商业和工业自动化生产领域中,嵌入式技术