资源描述:
《基于GMM区分性别的汉语方言识别系统.pdf》由会员上传分享,免费在线阅读,更多相关内容在应用文档-天天文库。
1、1002-8684(2011)12-0039-03基于GMM区分性别的汉语方言识别系统王侠a顾明亮a,b高原b马勇a徐州师范大学a.物理与电子工程学院;b.语言科学学院,江苏徐州221116[摘要]提出一种基于GMM的区分不同性别的汉语方言识别系统,系统提取语音的RASTA-PLP特征,在方言电话语音库上进行仿真实验,结果表明在GMM模型阶数为32时,系统的识别率可达到98.66%。同时还将RASTA-PLP特征与SDC特征对比,结果表明系统识别率最高可提高6.05%,且RASTA-PLP特征在性别分类方面优于SDC。方言识别;RASTA-PLP;GMM;SDCTN912.3AGMMBa
2、sedGenderDistinguishingandChineseDialectRecognitionSystemWANGXiaGUMingliangGAOYuanMAYong(3)4总结@@[1]VINCENTD.Torpedoangleandrangeestimationsusingmulti-pathssignalsandamodifiedfastorthogonalsearchalgorithm[D].Canada:RoyalMilitaryCollege,2008.@@[2]DONALDR,MCGAUGHEY,VINCENTD,etal.Improvedtorpedorange
3、estimationusingthefastorthogonalsearch[J].IEEEJournalofOceanicEngineering,2010,35(3):595-602.@@[3]吴百诗.大学物理(上册)[M].西安:西安交通大学出版社,2008.@@[4]朱代柱,惠俊英,蔡平.近程目标距离的实时估计研究[J].应用声学,1999,18(4):33-35.@@[5]蒋小勇,杜选民.强干扰背景下的鱼雷辐射噪声信号检测方法[J].声学技术,2010,29(1):18-22.2011-06-30der[C]//PrcceedingofIEEEInternationalSympo
4、siumon@@[1]ZISSMANMA.ComparisonoffourapproachestoautomatElectronicCommerceandSecurity.[S.1.]:IEEEPress,iclanguageidentificationoftelephonespeech[J].IEEE2010(1):84-88.Trans.SpeechandAudioProceeding,1996,4(1):@@[6]HERMANSKYH,MORGANN,BAYYAA,etal.RASTA-31-44.PLPspeechanalysistechnique[C]//Proceedingo
5、fIEEEIn@@[2]顾明亮,夏玉果,张长水,等.基于AdaBoost的汉语方言ternationalConferenceonAcoustics,Speech,andSignalPro辨识[J].东南大学学报:自然科学版,2008,38(4):cessing.[S.1.]:IEEEPress,1992(1):121-124.585-588.@@[7]KOHLERMA,KENNEDYM.Languageidentificationusing@@[3]HARBH,CHENL.Voice-basedgenderidentificationinshifteddeltacepstra[C]//P
6、roceedingofIEEEInternationalmultimediaapplications[J].InternationalJournalofPatternConferenceonDigitalSignalProcessing.[S.1.]:IEEERecognitionandArtificialIntellligence,2005,19(2):Press,2003(1):3-69.63-78.@@[8]ZHANGWQ,HEL,DENGY,etal.Time-frequencycep@@[4]邓英,欧贵文.基于HMM的性别识别[J].计算机工程与stralfeatureandh
7、eteroscedasticlineardiscriminantanalysis应用,2004,40(15):74-75.forlanguagerecognition[J].IEEETrans.Audio,Speech,@@[5]LIW,KIMDJ,KIMCH,etal.Voiced-basedrecognitionandLanguageProcessing,2011,19(2):266-276.systemfornon-seman