網(wǎng)絡(luò)音頻編碼算法研究和實現(xiàn).pdf

1、西安電子科技大學(xué)碩士學(xué)位論文網(wǎng)絡(luò)音頻編碼算法研究和實現(xiàn)姓名：宋少鵬申請學(xué)位級別：碩士專業(yè)：通信與信息系統(tǒng)指導(dǎo)教師：馬鴻飛20090101AbstractWiththerapiddevelopmentofIPbasednetworksandgreatincreaseofchannelbandwidths，multimediacontentiswidelyusedindailylifeandpeopleareexpectinghi曲一qua

2、lityaudiocommunicationsInthemeantime，thecomputingpoweranddevelopmenttoolsofDSPhavebeengreatlyimprovedandthepowerconsumptionandcostofDSPhavebeengreatlydecreased，S0theDSPiswidelyusedinrealtimemultimedia，includingaudiocommu

3、nicationsHowevertheIPbasednetworksarefacingpacketlossandbandwidthvariationwhichhavebadinfluenceonaudiotransmissionandisfarawayfromsolvedAccordingtothegeneralprinciplesofperceptualaudiocodingandMultipleDescriptionsCoding(

4、MDC)principles，anewnetworkaudiocodingalgorithmsisstudiedanddesignedinthispaperThemethodofdecompositionofaudiosignalsintohearingmaskingthresholdandresidualsignalsintermsofaudioanalysisandsynthesisisdiscussed，followedbyMDC

5、processingofbothhearingmaskingthresholdandresidualsignalsintermsofcodingandquantization，whichkeephighqualityaudioinreal—timecommunicationsofaudiosignalsThemethodsoftwodescriptionsscalarquantizingalgorithmstodealwithresid

6、ualsignalsisstudiedinthispaperThefixedpointimplementationofnetworkaudioalgorithmsusingtwodescriptionsscalarquantizingisimplementedontheTIDM642fixedpointDSEThedesignedalgorithmhasbeenevaluatedbyusingNS2networksimulatorThe

7、testingresultshaveshownthat，networkaudioalgorithmsusingtwodescriptionsscalarquantizingagainstpacketlossesandhaveabetterqualitythansingledescription，andshownthatthecyclecountsofthealgorithmsalereducedalotafteroptimization