语音自适应和差分编码方法.ppt

语音自适应和差分编码方法.ppt

ID:55666999

大小:10.20 MB

页数:43页

时间:2020-05-23

语音自适应和差分编码方法.ppt_第1页
语音自适应和差分编码方法.ppt_第2页
语音自适应和差分编码方法.ppt_第3页
语音自适应和差分编码方法.ppt_第4页
语音自适应和差分编码方法.ppt_第5页
资源描述:

《语音自适应和差分编码方法.ppt》由会员上传分享,免费在线阅读,更多相关内容在PPT专区-天天文库

1、SpeechSignalProcessing—Lecture7SpeechCodingMethodsBasedonSpeechWaveformRepresentationsandSpeechModels—AdaptiveandDifferentialCoding1QuantizationDilemmaWanttochoosequantizationstepsizelargeenoughtoaccommodatemaximumpeak-to-peakrangeofx(n);atthesametimen

2、eedtomakethequantizationstepsizesmallsoastominimizethequantizationerrorthenon-stationarynatureofspeech(variabilityacrosssounds,speakers,backgrounds)compoundsthisproblemgreatly2SolutionstoQuantizationDilemmaSolution1-letΔvarytomatchthevarianceoftheinput

3、signal=>Δ(n)Solution2-useavariablegain,G(n),followedbyafixedquantizerstepsize,Δ=>keepsignalvarianceofy(n)=G(n)x(n)constantCase1:Δ(n)proportionaltoσx2=>quantizationlevelsandrangeswouldbelinearlyscaledtomatchσx2=>needtoreliablyestimateσx2Case2:G(n)propor

4、tionalto1/σx2togiveσy2≈constantneedreliableestimateofσx2forbothtypesofadaptivequantizationAdaptiveQuantization:3TypesofAdaptiveQuantizationClassification1:instantaneous-amplitudechangesreflectsample-to-samplevariationsinx(n)=>rapidadaptationsyllabic-am

5、plitudechangesreflectsyllable-to-syllablevariationsinx(n)=>slowadaptationClassification2:feed-forward-adaptivequantizersthatestimateσx2fromx(n)itselffeedback-adaptivequantizersthatadaptthestepsize,Δ,onthebasisofthequantizedsignal,,(orequivalentlythecod

6、ewords,c(n))4FeedForwardAdaptationVariablestepsizeassumeuniformquantizerwithstepsizeΔ(n)x(n)isquantizedusingΔ(n)=>c(n)andΔ(n)needtobetransmittedtothedecoderifc’(n)=c(n)andΔ’(n)=Δ(n)=>noerrorsinchannel,andDon’thavex(n)atthedecodertoestimateΔ(n)=>needtot

7、ransmitΔ(n);thisisamajordrawbackoffeedforwardadaptation5Feed-ForwardQuantizertimevaryinggain,G(n)=>c(n)andG(n)needtobetransmittedtothedecoderCan’testimateG(n)atthedecoder=>ithastobetransmitted6FeedForwardQuantizersfeedforwardsystemsmakeestimatesofσx2,t

8、henmakeΔorthequantizationlevelsproportionaltoσx,orthegainisinverselyproportionaltoσx7FeedForwardQuantizertheparameterαcontrolstheeffectiveintervalofx(n)thatcontributestotheestimateofα=0.99=>bringsuplevelinlowamplituderegions=>syllabicra

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。