预览加载中,请您耐心等待几秒...
1/5
2/5
3/5
4/5
5/5

在线预览结束,喜欢就下载吧,查找使用更方便

如果您无法下载资料,请参考说明:

1、部分资料下载需要金币,请确保您的账户上有足够的金币

2、已购买过的文档,再次下载不重复扣费

3、资料包下载后请先用软件解压,在使用对应软件打开

基于AMR的语音质量提升方法研究 Abstract AutomaticSpeechRecognition(ASR)systemsarewidelyusedinvariousfieldssuchashealthcare,education,andfinance.ThequalityofASRoutputdependsonthequalityoftheaudioinput,whichmaybeaffectedbyvariousfactorssuchasnoise,codingartifacts,andchannelcharacteristics.Inthispaper,weproposeamethodtoimprovethequalityofAMR-encodedspeechusingvarioustechniquesincludingnoisereduction,de-artifacting,andchannelequalization.ExperimentalresultsshowthattheproposedmethodcansignificantlyimprovethespeechrecognitionaccuracyofAMR-encodedspeech. Introduction Speechrecognitionhasbecomeanessentialtechnologyinmodernsociety.However,thequalityofthespeechrecognitionsystemdependsonthequalityoftheinputspeech.Theacousticenvironmentmayintroducevariousnoises,distortions,andartifactsthatcandegradethequalityofspeechsignals,resultinginlowrecognitionaccuracy.Moreover,speechsignalsareusuallycompressedtoreducestorageandtransmissionoverheads.However,thecompressionprocesscanintroduceevenmoredistortionandcodingartifacts. AdaptiveMulti-Rate(AMR)isawidelyusedspeechcodingstandardthatcanprovidedecentspeechqualityatlowbitrates.AMRisusedinvariousapplicationssuchasmobilecommunication,videoconferencing,andvoicemessaging.However,AMR-encodedspeechsignalsaresusceptibletovariousdistortionssuchaschannelfading,noise,andcodingartifacts.Therefore,enhancingthespeechqualityofAMR-encodedspeechisessentialtoimprovetheperformanceofspeechrecognitionsystems. Inthispaper,weproposeamethodtoimprovethequalityofAMR-encodedspeechusingvarioustechniquesincludingnoisereduction,de-artifacting,andchannelequalization.TheproposedmethodcanbeusedtoenhancetheperformanceofbothspeechrecognitionandspeakerverificationsystemsthatoperateonAMR-encodedspeechsignals. Relatedwork Varioustechniqueshavebeenproposedtoimprovethequalityofspeechsignals.Forexample,noisereductiontechniquescanremovetheadditivenoisefromspeechsignals,thusimprovingtheperceptualqualityofspeech.De-artifactingtechniquescanremovethecodingartifactsintroducedbycompressionandtransm