基于深度神经网络的单通道语音增强算法研究与实现

《基于深度神经网络的单通道语音增强算法研究与实现》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、分类号TP391.9学号14063057UDC密级公开工程硕士学位论文基于深度神经网络的单通道语音增强算法研究与实现硕士生姓名孙海霞工程领域计算机技术研究方向数字媒体语音信号处理指导教师李思昆教授国防科学技术大学研究生院二〇一六年一月ResearchandImplementationofSingle-channelSpeechEnhancementBasedonDeepNeuralNetworkCandidate：SunHaiXiaAdvisor：Prof.LiSiKunAthesisSubmittedinpartialfulfillmentofth

2、erequirementsfortheprofessionaldegreeofMasterofEngineeringinComputerTechnologyGraduateSchoolofNationalUniversityofDefenseTechnologyChangsha，Hunan，P.R.China（January，2017）国防科学技术大学研究生院工程硕士学位论文目录摘要.....................................................................................

3、............................iABSTRACT........................................................................................................iii第一章绪论.................................................................................................11.1研究背景和研究意义...................

4、..............................................................11.2国内外研究现状.........................................................................................21.2.1传统语音增强算法的发展..................................................................21.2.2有监督语音增强技术发展................

5、..................................................31.2.3深度神经网络语音增强技术发展......................................................71.3本文组织结构.............................................................................................81.4本文研究成果.........................................

6、....................................................9第二章基于深度神经网络的语音增强算法原理..........................................112.1语音模型...................................................................................................112.1.1噪声分类............................................

7、................................................112.1.2语音质量评价....................................................................................112.1.2.1PESQ原理...................................................................................................122.1.2.2分段信噪比（SegS

8、NR）...................................................