Support User Manuals

Telos Zephyr Xstream Two-Way Radio User Manual

Open as PDF

of 381

USER’S MANUAL

Section 6: AUDIO CODING REFERENCE 110

delta,betweensuccessiveaudiosamplescomparedtousingtheindividualvalues.Further

efficiencyishadbyadaptivelyvaryingthedifferencecomparatoraccordingtothenatureofthe

programmaterial.G.722andAPT‐XareexamplesofADPCMschemes.Theyachievearounda

factorof4:1reductioninbitrate.

G.722achievesadditionalefficien

cybyallocatingitsbitstomatchthepatternsinthehuman

voice,andit’sconsideredadequatefornewsandtalkprogrammingoverISDN.But,forhigh‐

fidelitytransmission,

algorithmswithmorepowerarerequired.Thesearebasedon

psychoacoustics,wherethecodingprocessisadaptedtothewaywehearsounds.Thereare

severalalgorithmsavailable,withvaryingcomplexityandperformancelevels.

Someyearsago,theinternationalstandardsgroupISO/IECestablishedtheISO/MPEG(Moving

PicturesExpertGroup),todevelopauniversalstandardforen

codingmovingpicturesandsound

fordigitalstorageandtransmissionmedia.ThestandardwasfinalizedinNovember1992with

threerelatedalgorithms,calledLayers,definedtotakeadvantageofpsychoacousticeffects

whencodingaudio.Layer1and2areintendedforcompressionfactorsofabout4:1and6or

8:1respectively,andthes

ealgorithmshavebecomepopularinsatelliteandhard‐disksystems.

Layer‐3achievescompressionupto12.5:1—8%oftheoriginalsize—makingitidealforISDN.



Basic Principles of Perceptual Coding

Withperceptualcoding,onlyinformationthatcanbeperceivedbythehumanauditorysystemis

retained.

Lossless–which,foraudio,translatestonoiseless–codingwithpe

rfectreconstructionwould

beanoptimumsystem,sincenoinformationwouldbelostoraltered.Itmightseemthat

lossless,redundancy‐reducingmethods(suchasPKZIP,Stuffit,Stacker,andothersusedfor

computerhard‐diskcompression)wouldbeappli

cabletoaudio.Unfortunately,noconstant

compressionrateispossibleduetosignal‐dependentvariationsinredundancy.Therearehighly

redundantsignalslikeconstantsinetones(wheretheonlyinformationnecessaryisthe

frequency,phase,amplitude,anddurationofthetone),whileothersignals,suchasthosewhich

approachbroadbandnoise,maybecompletelyun

predictableandcontainnoredundancyatall.

Furthermore,lookingforredundancycantaketime.Whileapopularsongmighthavethree

choruseswithidenticalaudiodatathatwouldneedtobecodedonlyonce,you’dhavetostore

andanalyzetheentiresonginordertofi

ndthem.Anysystemintendedforareal‐timeuseover

telephonechannelsmusthaveaconsistentoutputrateandbeabletoaccommodatetheworst

case,soeffectiveaudiocompressionisimpossiblewithredundancyreductionalone.

Fortunately,psychoacousticspermitsacleversolution!Effectscalled“masking”havebeen

discoveredinthehu

manauditorysystem.Thesemaskingeffects(whichmerelyprovethatour

brainisalsodoingsomethingsimilartobitratereduction)havebeenfoundtooccurinboththe

frequencyandtimedomainsandcanbeexploitedforaudiodatareduction.

Mostimportantforaudiocodingaretheeffectsinth

efrequencydomain.Researchinto

perceptionhasrevealedthatatoneornarrow‐bandnoiseatacertainfrequencyinhibitsthe

audibilityofothersignalsthatfallbelowathresholdcurvecenteredonamaskingsignal.

Thefigurebelowshowstwo“thresholdofaudibility”curves.Theloweroneisthetypical

frequen

cysensitivityofthehumanearwhenpresentedwithasingleswepttone.Whenasingle

constanttoneisadded,thethresholdofaudibilitychangesasshownintheuppercurve.The

previous next