盲信号处理毕业论文

学院毕业设计1学院毕业论文设计盲信号处理院（系）：电子信息系专业：电程技术班级：班学号：030姓名：指导教师：学院毕业设计2目录摘要..............................................................................................................................4ABSTRACT...............................................................................................................5第一章语音信号及噪声概述.............................................................................11.1语音信号的概述....................................................................................................11.1.1语音特性分析......................................................................................11.1.2语音信号的基本特征..........................................................................21.2语音噪声特性分析..................................................................................................31.2.1信噪比(SignalNoiseRatio，SNR).....................................................31.2.2信干比(signal-to-InterferenceRatio，SIR)........................................4第二章盲信号处理................................................................................................52.1盲信号处理的概述.................................................................................................52.1.1盲信号处理的基本概念......................................................................52.1.2盲信号处理的方法和分类....................................................................62.1.3盲信号处理技术的研究应用................................................................62.2盲源分离法.............................................................................................................72.2.1盲源分离技术........................................................................................72.2.2盲分离算法实现....................................................................................72.2.3盲源分离技术的研究发展和应用........................................................82.3独立成分分析........................................................................................................92.3.1独立成分分析的定义............................................................................92.3.2ICA的基本原理...................................................................................102.3.3本文对ICA的研究目的及实现.........................................................12第三章盲语音信号分离的实现及抑噪分析...............................................153.1盲语音信号分离的实现......................................................................................153.1.1盲信号分离的三种算法......................................................................153.1.2不同算法的分离性能比较..................................................................17学院毕业设计33.2抑制噪声的算法仿真及结果分析....................................................................173.2.1抑噪算法仿真实现..............................................................................173.2.2分离结果分析.......................................................................................203.2.2不同算法的分离性能比较..................................................................28第四章结论与展望..............................................................................................35致谢.....................................................................................错误!未定义书签。参考文献...................................................................................................................37附录...................................................................................................................38学院毕业设计4盲信号处理摘要语音信号盲分离处理的含义是指利用BSS技术对麦克风检测到的一段语音信号进行处理。混合语音信号的分离是盲分离的重要内容，目前的混叠语音分离大多是建立在无噪环境中的混叠情形下，主要以盲源分离(BlindSourceSeparation，BSS)，根据信号的统计特性从几个观测信号中恢复出未知的独立源成分。本文重点研究了以语音信号为背景的盲处理方法，在语音和听觉信号处理领域中，如何从混有噪声的的混叠语音信号中分离出各个语音源信号，来模仿人类的语音分离能力，成为一个重要的研究问题。具体实现主要结合ICA技术，将语音去噪作为一个预处理过程，对带噪声的混叠语音盲分离进行了研究，本文详细了介绍三种FastICA算法：SOBI算法以及CICA算法，将三种算法应用于实际的语音信号噪声分离中，并比较了传统算法和基于ICA/BSS算法在语音消噪和增强方面的差异，文章最后还介绍了分离效果评价准则，并比较了SNR和SIR参数。关键词：语音信号，语音信号噪声，盲源分离，独立成分分析学院毕业设计5ABSTRACTBlindseparationofspeechsignalprocessingmeansistheuseofBSStechniquesmicrophonedetectsavoicesignalprocessing.Separationofmixedspeechsignalsisimportantforblindseparation,thecurrentseparationofoverlappingspeechisbuiltmostlyintheabsenceofaliasingnoiseintheenvironment,underthecircumstances,mainlyinblindsourceseparation(BlindSourceSeparation,BSS),thestatisticalcharacteristicsofsignalSeveralobservationsfromtheunknownsignaltorecovertheindependentsourcecomponents;Thisarticlefocusesonthebackgroundtotheblindspeechsignalprocessingmethod,signalprocessing,speechandhearing,howthenoisefromamixtureofoverlappingspeechsignaltoseparatethevariousaudiosourcesignalstomimichumanspeechseparationabilitybecomeanimportantresearchquestion.CombinationofconcreterealizationofthemainICAtechnology,speechdenoisingasapreprocessingofspeechwithnoise-aliasingblindseparationwerestudied,thepaperintroducesthreekindsofFastICAalgorithm:SOBIalgorithmandtheCICAalgorithm,threekindsofalgorithmapplicationtheactualnoiseinspeechsignalseparation,andcomparedthetraditionalalgorithmandtheICA/BSSalgorithminspeechdenoisingandenhancementofthedifferences,thepaperfinallydescribestheseparationevaluationcriteria,andc