全部 标题 作者
关键词 摘要

OALib Journal期刊
ISSN: 2333-9721
费用:99美元

查看量下载量

相关文章

更多...

基于emd的acf基音检测改进算法

, PP. 42-47

Keywords: 基音,经验模式分解,自相关函数,本征模式函数

Full-Text   Cite this paper   Add to My Lib

Abstract:

针对传统的自相关函数基音检测算法容易出现倍频错误的问题,本文提出了一种基于经验模式分解的acf基音检测改进算法.该改进算法利用emd将一帧语音信号的acf分解成多个本征模式函数和残余分量,同时根据imf的累积能量分布情况找出含有基音信息的imf,最后通过该imf准确地估计出该语音帧的基音.仿真实验结果表明:本文所提算法性能明显优于传统acf算法;相比较于检测效果较好的wac算法,本文所提算法的性能依然有了一定的提升.

References

[1]  shimamurat,kobayashih.weightedautocorrelationforpitchextractionofnoisyspeech[j].ieeetransactionsonspeechandaudioprocessing,2001,9(7):727-730.
[2]  rabinerlr,chengmj,mcgonegalca.acomparativeperformancestudyofseveralpitchdetectionalgorithms[j].ieeetransactionsonacoustics,speechandsignalprocessing,1976,24(5):399-417.
[3]  rossm,shafferh,freudbergr,etal.averagemagnitudedifferencefunctionpitchextractor[j].ieeetransactionsonacoustics,speechandsignalprocessing,1974,22(5):353-362.
[4]  kadames,broudreaux-bartelsgf.applicationofthewavelettransformforpitchdetectionofspeechsignals[j].ieeetransactionsoninformationtheory,1992,38(2):917-924.
[5]  amadog.pitchdetectionalgorithmsbasedonzero-crossrateandautocorrelationfunctionformusicalnotes[c]//proceedingsoficalip.shanghai:ieee,2008:449-454.
[6]  hasanmk,hussains,setumth,etal.signalreshapingusingdominantharmonicforpitchestimationofnoisyspeech[j].signalprocessing,2005,86(5):1010-1018.
[7]  huangne,zhengs,longsr,etal.theempiricalmodedecompositionandthehilbertspectrumfornonlinearandnon-stationarytimeseriesanalysis[c]//proceedingsofroyalsocietya.london:royalsociety,1998:903-995.
[8]  mollam,khademuli,hirosek,etal.pitchestimationofnoisyspeechsignalsusingempiricalmodedecomposition[c]//proceedingsofinterspeech.antwerp:isca,2007:2117-2180.
[9]  meyerg,plantef,ainsworthwa.apitchextractionreferencedatabase[c]//proceedingsofeurospeech.madrid:isca,1995:827-840.
[10]  ahmadis,spaniasas.cepsrum-basedpitchdetectionusinganewstatisticalv/uvclassificationalgorithm[j].ieeetransactionsonspeechandaudioprocessing,1999,7(3):333-338.

Full-Text

Contact Us

service@oalib.com

QQ:3279437679

WhatsApp +8615387084133