AudibilityoftemporalsmearingandtimemisalignmentofacousticsignalsMilindNKunchurDepartmentofPhysicsandAstronomyUniversityofSouthCarolinaColumbiaSC29208Dated19072007received29082007published ID: 316075
Download Pdf The PPT/PDF document "TechnicalAcoustics,http://www.ejta.org,2..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
TechnicalAcoustics,http://www.ejta.org,2007,17.CopyrightTechnicalAcoustics[ISSN1819-2408](2007)EEAA. AudibilityoftemporalsmearingandtimemisalignmentofacousticsignalsMilindN.KunchurDepartmentofPhysicsandAstronomyUniversityofSouthCarolina,Columbia,SC29208(Dated:19.07.2007[received];29.08.2007[published])Misalignmentintimingbetweendriversinaspeakersystemandtemporalsmearingofsignalsincomponentsandcableshavelongbeenallegedtocausedegradationofdelityinaudioreproduction. reproductionsincesuchdistortionscannotnaturallyariseinanaudiochain.Thepresentworkreportsthedetectionofatemporalde-layshorterthananythathasbeenpreviouslypublished,andonewhosethresholdunderceeds1.Further-more,themethodandtypeofthetemporalfeatureadisparitybetweenspatialpathdistancesfromtwoloud-speakerdriversisadistortionthatcanactuallybeman-ifestedinareal-lifeaudiosetup,sincemostspeakersys-temsconsistofmultipledriversandeventhosethatdontwillexhibitatemporalspreadbecauseofthenitedi-mensionsofthedriver.Sucharrival-timediscrepanciescanplayanevengreatersignicanceinmulti-channelsurround-soundsystems.Thepresentresultprovidesascienticbasisfortheanecdotalclaimsbyaudiophilesthatdelityrequirestimeresponseinthemicrosecondrange,andprovidesasolidquantitativestandardforas-sessingthedeterioratingeectsoftemporaldelaysandsmearinginanaudiochain.1BACKGROUNDThecentralgoalinhigh-delitysoundreproductionistoreproduceasoundwithsucientaccuracysuchthattheerrorsinalldomainsarebelowtheirthresholdsofde-tectability.Settingasidestereoandspatial-localizationaspects,monauralsoundscanbeperceivedasdierentbecauseof(a)dierentfrequencycomponents,(b)dif-ferentlevelsofcomponents,(c)dierentrelativetimingsandrisetimesofcomponents,and(d)dierentphasesoftheindividualcomponents.Thersttwodierencesareoftencollectivelyreferredtoasspectral;although,strictlyspeaking,theseshouldbereferedtoasdierencesintheamplitudespectrumorintensityspectrum,sinceallalterations,includingthoserelatedtophaseandtime,canbedescribedthroughchangesinthecomplexFourierspectrum.Exceptinsimplelinearsystems,theinter-relationsbetweenfrequency,amplitude,time,andphasearenotstraighforward.Inaudiosystems,forex-ample,acrossovercanintroduceafrequencydependentphasedierencewithoutphysicallydelayingtheonsetofonefrequencybandwithrespecttoanother;however,theeectofhavingunequallistener-to-subwooferandlistener-to-satellitespeakerdistancesisbestdescribedbyanoveralldelaybetweenthetwofrequencybands.Similarly,thehearingmechanismtreatsphaseandtimedierencesonseparatefootingsanderrorsinthetwodonothaveequivalentconsequences.Inalinearcircuit,adelayinasinusoidalsignalisrelatedtoitsphaseshift.Inthehearingprocess,thesoundsignalisdecomposedintoseparatefrequencychannelsthroughanarrayofsensoryinnerhaircells(IHCs)tunedtodierentcharacteristicfrequencies(CFs)andarrangedtonotopicallyalongthebasilarmembraneinthecochlea.Thistonotopicallyseparatedinformationiscarriedbyau-ditorynervebers(ANFs)tothecochlearnucleus(CN).ThenerveimpulsesalongtheANFsfollowthephaseoftheircorrespondingacousticsignalsforfrequenciesuptoabout4kHz.However,thehearingmechansimlargelyabandonscross-frequencyphaseinformationleadingtothefamousOhms(second)law[5,6]wherebytheearisnotacutelysensitivetophaseshiftsbetweenwellsepa-ratedfrequencies(despitelargedierencesinwaveformshape).Thisfactishelpfulinthedesignoffrequency-cross-overcircuitswherephasedierencesbetweenlow-passandhigh-passoutputsareimportantmainlytotheextentthattheyaecttheamplituderesponse[3].Above4kHz,theANFsrespondapproximatelywithaplateauofactivityforthedurationofthetonewithnosynchro-nizationbetweentheringpatternandthephaseoftheacousticsignal.Whilecross-frequencyphasecoherenceislessimportant,timecoherenceisadierentmatter.Theauditorysys-temisverysensitivetothesynchronicityintheonsetsofdierentfrequenciesasiswellknown,instrumentaltimbrebecomesambiguousiftheonsetsanddecaysofthenotesareremoved[7].Inthecochlearnucleus,fastrespondingoctopuscellsactassynchronousANDgates60inputseach)toconvergecoincidentANFsig-nalsfromdierentfrequencychannels[8,9].Thesecellsrespondsharplytowelltimedmultichromaticactivityattheonsetofasoundandthustheiroutputisagaugeofthestimulusslewrate[10].IftheinitialtemporaluncertainityintheANFsignalisrepresentedbyaGaus-sianprobability-densityfunction,thenscanbetakenasaroughestimateoftheini-tialtemporalspreadsincetheANFslosephaselockingwiththeacousticstimulusaround4kHzandrespondonlytothepositivehalfcycle[11,12].Theprobabil-ityforNsignalstoarrivesimultaneouslyinordertoexciteanoctopuscellisproportionaltotheproductoftheprobabilities.Thisgivesanoutputprobabilityfunc- withareducedtemporalspreadof .BesidestheinitialconvergencefactorofN60ateachoctopuscell,theoctopus-celloutputsundergoadditionalconvergenciesathigherneurallevels(e.g.,insphericalbushycellsinthelateralleminiscus).ItisnotcleartowhatextenttheseconvergenciesboostNandimprovethetimeresolution;however,itisclearthatthemaximumpossibleconvergencefactorcannotex-ceedthetotalnumberofIHCs,whichis4000.Thusfortransientstimuli,theauditorysystemstemporalacuitymaybeestimatedtobeinthe2 16srange,taking withN=60 4000.Noticethatthevalueofthishasverylittletodowiththehigh-frequencyaudibil-itylimit.Asdescribedearlier,theauditorysignalsoriginatefromhaircellsarrangedsuchthattheonesclos- 4 d FIG.1:Experimentalconguration.Speaker-to-listenerdis-tance=4.3m,aperturelength=1.5cm,andspeaker-centertospeaker-centerdistancecm.Misalignmentisvariable.Duringblindtrials,alistenertriestodistinguishbetweenthealigned(=0)andmisaligned(settingsforvaluesranging2 10mm(2METHODS2.1ApparatusThecongurationoftheexperimentisshowninFig.1.Twoloudspeakersarestackedverticallyontopofeachotherwiththeirfrontfacesparalleltoeachother.Thetopspeakerismountedonrailsandcanslidebackandforthbetweenaxedstop(forthealignedposition)andamicrometer-setscrewadjustablestop(forthedisplacedposition)throughasetdisplacement.Thelistenerisseatedatadistance3m,facingthespeakerswithearsataheightmidwaybetweenthetwospeakers.Thespeakersarelaterallycenteredw.r.t.(withrespectto)bothearssothatbothearsreceivethesamesignal.Theroomshapeisarectangularparallelepipedwithaheightof2.7m,awidthof3.6m,andalengthof5.8m.Thespeaker-listeneraxisliesalongthelongdimensionandiscenteredw.r.t.thesidewalls;thisaxisisataheightof1.1mabovetheoor.Theoorandwallsoftheroomwerecoveredwithacousticalcarpetingandtheceilingcoveredwithacousticaltiles.Thesematerialshaveabsorptioncoecientsof7atthefrequenciesofinterest(7kHz).Inaddition,panelsmadefrom38mmthickglass-berboards(forwhich95)wereplacedatcertainstrategiclocationstosuppressprincipalreections(thereweresixsuchpanelswithatotalareaofabout9mTheloudspeakersusedwereapairofAurumCantusG2Siribbontweeters(JinlangAudioCo.Ltd.,PenglaiCity,P.R.ofChina)whichhaveafrequencyresponseof2 40kHz,asensitivityof96dB/Wat1m,andanom-inalimpedanceof6.Bothspeakerswereconnectedinparalleltothesame7kHzsquare-wavesignalsource.Thissignalsourceconsistedofananalogsignalgenerator(model4001manufacturedbyGlobalSpecialtiesInstru-ments,Cheshire,Connecticutt)followedbyawidebandamplier(witha3-dBpowerbandwidthof0 2.2MHz).Fig.2showsthevoltagewaveformatonespeakersinputterminalsmeasuredwithaLeCroymodelLT322(LeCroyCorporation,ChestnutRidge,NewYork)500MHzdig-italstorageoscilloscope,whichdigitizedthesignalatasamplingrateof200MS/s(millionsamplespersecond)anda12-bitverticalresolution;thissameoscilloscopewasusedinalltheotherwaveformandspectrummea-surements.Notethewellcontrolledresponsewithneg-ligibleringingandovershoot,andrise/falltimesofs.Themeasuredjitterinthissquare-wavesignalwas68ns(05%oftheperiod). FIG.2:Waveformofvoltageatloudspeakerterminals,recordedata200MS/ssamplingrate.A7kHzsquarewaveformwaschosenbecauseithasonlyoddharmonicsthat,otherthanthefundamental,arebelowtheirrespectivesingle-toneaudibilitythresholds.However,theacousticoutputfromaloudspeakerwillnotbeaperfectreplicaofitselectricalinput.Besidesalteringtheharmoniccoecientsbecauseofanincon-stantfrequencyresponse,subharmonicsandotherspuri-ousanharmoniccomponentsmaybegenerated[35]whenaspeakerisdrivenathighlevels,especiallywithinad-equatedamping.Inthepresentexperimentthedrivinglevelismodest(0.5Vpeakinputvoltageand69dBSPLsoundlevelatlistenerposition)andthedampingiseective(40msignal-sourceoutputresistance)topre-ventanharmonicdistortion.Thisabsenceofanharmonicdistortionwasveriedbyspectrumanalyzingtheacous-ticoutputoftheloudspeakerusinganACOPacic(ACOPacic,Inc.,Belmont,California)model7016measure-mentmicrophoneanda4012preamplierwitha40dBgainstage.Thefrequencyresponseofthemicrophonetogetherwithitspreamplierwasat(3dB)withina4Hz 120kHzband.ThispowerspectrumoftheacousticsignalfromtheloudspeakerisshowninFig.3(a).Panel(b)showsamagniedviewofthefundamental()peakwithlinearaxes;thefull-widthhalfmaximum(FWHM)is0.77Hz0.01%of).Nosubharmonicpeakscouldbedistin-guishedfromnoise.Theabsolutesoundlevelofthenoise(includinginthe3.5kHzsubharmonicvicinity)wasdBSPL(i.e.,belowthedashedlineinFig.3[a]).Therewasalsonodetectableanharmoniccontentabove.Asexpected,thespectrumisdominatedbyoddharmonics(7,21,35,49kHz,...)whichextendwellintotheultra-sonicrange.Becausetheentiresignalchainisanalog,spuriousfrequenciesthatcanresultfromaliasingindig- italsystemswereavoided. FIG.3:Powerspectrumofthe(unaveraged)acousticoutputofoneloudspeakeratadistanceof0.7m.Thepowercoecientsarenormalizedw.r.t.thefundamentalpeak.(a)Log-linearplotofthe20Hz 80kHzwindowin20Hzstepstakenata2MS/ssamplingrate.Thehorizontaldashedlinecorrespondstotheabsolutesoundlevelof0dBSPL.(b)Linear-linearplotoftheregionnearthefundamentalpeakin0.5Hzstepstakenata50kS/ssamplingrate.2.2AcousticstimuliatthelistenerpositionReferringbacktoFig.1,thetwosignalsarriveatthelistenersearswithaprimaryrelativedelayofisthespeedofsound.Asmentionedearlier,thelistenerisseatedatadistance3m,facingthespeakerswithearsataheightmidwaybetweenthetwospeakers.Thealignmentoftheapparatusandlistenerpositionshouldbesuchthatthelinejoiningthemidpointbetweenthelistenersearsandthemidpointbetweenthetwospeakersshouldbeperpendiculartotheplanedenedbythespeakerfrontsurfaceswhentheyareundisplaced.Thiswasrstcheckedwithalaserbeamandthenthelistenerfurthernetunedhis/herheadcenteringbyear.Fromthispointonwardthelistenerheldhis/herheadstillinthisxedposition.Thesoundlevelatthesubjectlocationwas69dBSPLforEq.1wouldbeexactandcompleteifthetwospeakerswerepointsourcesandtherewerenoroomreections.However,soundproducedbyanextendedsourcewillsuf-feranintrinsictemporalspread(becausewaveletsem-anatingfromdierentpointsoftheradiatingsurfacewillarriveattheearsatdierenttimes)whichtendstoreducethesensitivityoftheexperiment.Heresoundisgener-atedineachtweeterbya6cmlongverticalaluminumribbon.AsshowninFig.1,absorbentbaeswereusedtocuttheaperture(andeectiveribbonlength)tocm(with=4.2cm),thusreducingthetemporalspreadwhichissmallcomparedwiththedelaysprobedintheexperiment.Theverticalseparationof9cm,betweenthecentersoftheupperandlowerspeakers,re-sultsinanangularseparationbetweenthesesourcesofatthelistenerposition.Aswiththetemporalspreadswithinthesources,roomreectionscanalsodiminishthetemporaldenitionofthedelay.Whileprecautionsweretakentominimizereectedenergy,eventhebestanechoicchamberwillnothaveperfectabsorption.Thereforeitisnecessarytoquantitativelyassesstheeectofreectedenergyontheexperiment.Thesignalatthelistenerpositionwillconsistofthedirectradiationfrombothspeakersplusthesumofallreections.Everyreectedpathoriginat-ingfromonespeakercanbeassociatedwithamatchingreectedpathoriginatingfromtheotherspeakerandallreectionscanthusbeconsideredinsuchmatchedpairs.Labellingeachsuchreectionpairby=0cor-respondingtothepairofdirectsignals),aspeakermis-alignmentintroducesanincrementalpathdierencebetweenthemembersofapair.Itisobviousfromthege-ometrythatinallcases.Infactfortherstreec-tionsfromthesidewalls,dD/ istheroomwidthtransversetothespeaker-listeneraxis)andfortherstreectionsfromtheoorandceiling,dD/ +(2istheverticaldistancefromthespeaker-listeneraxistotheoororceil-ing).ThesescorrespondtodelaysThedirectsignalandthereectionfromthebackwall(behindthelistener)bothhaveandcorrespondexactlytotheprimarydelayof.Higher-orderreec-tionswillhaveprogressivelydiminishingdelays.Besideshavingashorterincrementaldelaybetweenthetwospeakersignals,apairofreectedsignalscanalsohaveanextrainitialgeometricalpathdierencebe-tweenpairmembers.Forpathsthatundergoasinglereectionfromeithertheoorortheceiling, D22 D2 D2 andforpathsthatundergoasinglereectionfromany ),where7kHz.ThesecoecientsaregiveninTableIIandhavebeennor-malizedw.r.t.therst-harmoniccoecientofthecontrolwaveform(i.e.,by[0]).Columnsforharmonicswherevaluesfallwellbelowthenoiseoor(0.005)havebeenexcluded(thisisthecaseforallevenharmonicsex-ceptfor=2,whichisontheborderline).Thephaseofeachharmonicisspeciedrelativetothephaseofthefundamentalforthatsamevalue(i.e.,w.r.t.w.r.t.d]);theabsolutephaseandthephasedierenceacrossdierentvaluesofareofcourseinconsequential.Besidesthenoiseerror,thevaluesofthenon-zerocoecientsvarytosomeextentdependingonthepositionofthemicrophonebecauseofsmalllocalvariationsinintensitycausedbyweakpartialstandingwaves.Thispositionalsensitivityproducesanuncertainityofabout015inthenon-zerocoecients,whichisequivalenttoanerrorinthemeasuredattenuationsinthe13to16dBrange.Inaperfectsquarewaveform,theharmoniccoecientsaregivenbyforodd=0foreven.ItcanbeseenfromTableIIthattheevenindeednegligibleandthattheoddfollow1wellupto=3.Beyondthatthehigh-frequencyfalloofthespeakersresponseisevidentandthecoecientsaresmallerthantheir1theoreticalvalues.Thetotalsoundlevelat=0is69dBSPL.Most(88%)ofthispoweriscontainedinthe7kHzfundamental.Thelev-elsofallharmonicsbeyond7kHzatall(forexampleexample12dBandand60dB)fallbelowtheirthresholdsofaudibility[35 37].ThelastTABLEII:Harmoniccontentsofacousticsignals.Coe-areexpressedasafractionof.Phases,inradians,areexpressedrelativetotheforthesamevalue.Thelasttwolowercolumnsgivethepowerattenu-ations,indB,inthetotalrmsstrengthsandrst-harmoniccomponents()relativetotheirundisplaced()con-trolvalues.Thenoiseooranderrorbarforthecoecientsare0.005and0.015respectively. =7kHz =14kHz =21kHz =35kHz 1 1 C2 2 C3 3 C5 5 0 0.00 0.006 2.9 0.00 0.006 6.2 0.00 0.005 0.816 0.00 0.004 =49kHz =63kHz =77kHz Attenuation 7 7 C9 9 C 1 0 0 0.24 0.60 10.3 0.007 1.76 twolowercolumnsinTableIIgivetheattenuationsinthetotalrmspower()andinthefundamentalcomponent(presumablytheonlyaudi-bleFouriercomponent)relativetotheirvaluesfor=0.Thesemeasuredattenuationsinthefundamentalcom-ponentareplottedinFig.5andareseentomore-or-lessagreewiththetheoreticalcurveofEq.7andcorrespond-ingcalculatedupperboundsofTableI. FIG.5:Powerattenuationofthe7kHzfundamentalcom-ponentasafunctionofspeakerdisplacement.Thesymbolsshowthemeasuredattenuationintheacousticsignalatthelistenerposition.ThesolidlineshowsthetheoreticalcurvecorrespondingtoEq.7.Comparedtotheabovemeasuredacousticwaveforminair,thewaveformattheeardrumwillbepoorerinhigher-harmonicamplitudesbecauseoflteringbytheearcanal.So,forexample,thethird-harmonictorst-harmonicratiosattheeardrumwillbelowerthanthemeasured)ratiosofTableII.However,thefrac-tionalchangeineachFourieramplitudeattheeardrumwillbeexactlythesameasthecorrespondingmeasured(0)ratioinTableII.Thustheattenuationsinthefundamentalattheeardrumwillbeexactlythesameasthemeasuredvaluesgiveninthelastlowercolumnofthetableandthetotalrmsattenuationattheeardrumwillbemarginallylowerthanthesecond-lastlowercol-umnofthetable.2.3ProcedureInthisexperiment,subjectsareseatedinfrontofthetwocloselyspacedspeakers(Fig.1).Forthecontrolcondi-tion,thespeakerswerealignedandequidistantfromthelistenersears;forthetestcondition,thespeakersweremisalignedbyadisplacement.Theacousticwaveformattheearsbecomesprogressivelytemporallysmeared(Fig.4andEq.6)andtheharmonicsincreasinglyat-tenuated(Eq.7)asisincreased.Thecontrolsound=0)wasperceivedtohaveasharperorbrightertim-brethanthedisplacedsetting(=0),untilbecametoosmalltomakeadierence.Thegoalwastondthethatcouldbarelybediscriminated.Intheblindtest,thesubjecttriestojudgewhetheranunknownsoundcorrespondstothecontrolordisplacedsettingfordierentvaluesof.Itwasfoundthatsubjectstypicallyneedtolistentothesoundsforafewsecondstoforma 9 FIG.6:Summaryofresultsasafunctionofthetimede-lay(averagedacrossallsubjects).Eachdatapointconsistsof50blindtrials.(a)Thepercentageofcorrectjudgements.(b)Chi-squaredvalue.Thedashedline,correspondingtothecriticalvalueof3.84,intersectsthedatacurvearoundbetweenmultipledrivers)introducetemporalsmearinganddelays.Inthelastcase,thetemporalsmearingcanbeenormous.Forexample,Eq.2indicatesthatadipoleloudspeakerwithasingleelectrostaticpanelofheight=1.5mataspeaker-listenerdistanceof=5m(withthelistenersearathalfspeakerheight)willhaveatem-poralspreadof65ms.Whatthismeansisthateveniftheentirechainhadanotherwiseunlim-itedbandwidth,adelta-function(narrowimpulse)inputsignalwillgetspreadoutovera650slongrectangu-larwindowatthelistenerposition.Thusaloudspeakerthatsubtendsalargeangleatthelistenerpositionmustnecessarilycompromisedelity,perhapsexplainingwhysmallspeakerstendtohaveasubjectivelycleanerandmorecoherentsound(althoughtheymaybedecientintheirlow-frequencyresponse).Thepresentworkprovidesthebestcurrentquantitativeassessment(s)astowhatextentsuchtemporalerrorsmakeanaudibledierence.Thevastmajorityofpreviouspsychoacousticexperiments(summarizedintheBackgroundsection)thatprobedthisquestion,usedequipmentwhoseownowntemporalresponsemayhavebeenamajorlimitation.Mostofthatresearchusedrathercoarsedigitalsynthesisforthesignalsource,usedamplicationofinsucientintrinsicresponsespeed,andtransducerswithlimitedbandwidthsthatweredrivenwithinadequatedamping.Thepresentworkusesanana-logchaininwhichthesquare-wavesignalpresentedtothetransducer(includingtheresponseofboththesignalgen-eratorandamplier)hasrise/falltimesthatareabout100timesfasterthan48-kHzsampling-ratedigitalsyn-thesis.Theunsuallylow(40m)outputimpedancethatsourcesthetransducersprovidesexceptionaldampingandconsequentlyawellcontrolledwaveform,asshowninFig.2.Thetransducersusedinthisworkhaveafarmoreextendedbandwidth(spectrumshowninFig.3)comparedwithtypicaltransducersusedinaudiometry(e.g.,TDH-39headphones).Thusbyliftingsomeoftheequipmentbottlenecks,itwaspossibletodemonstrateamuchshorterthresholdfordiscerningtemporalerrors,thanhasbeenachievedbefore(asnotedintheIntro-ductionsection,thedenedinreference[31]corre-spondstoaninterpulsedelayof=10+thatismuchlongerthanthethresholdobtainedhere).Thisnewlowerthresholdshouldbetakenintoaccountinthedesignandsetupofaudiocomponentsifthehighesttransparencyistobeachieved.Whilethepresentdemonstrationofdiscriminabilityatthemicrosecondtimescaleusedsimple(square-waveform)high-bandwidthsignals,realisticmusicalsoundsalsocarrycontentinthistemporal-spectralrange.Measurementsofspectraofvariousmusicalinstrumentsshowthattheseextendintotheultrasonicrange[38]andevenbeyond100kHz[39].Inthetimedomain,ithasbeendemonstratedthatseveralinstruments(xylophone,trumpet,snaredrum,andcymbals)haveextremelysteeponsetssuchthattheirfullsignallevels,exceeding120dBSPL,areattainedinunder10s[2,38].Besidesultrasonicspectralcontentandmicrosecond-rangeonsetdurations,athirdaspectofmusicalsoundthatdemandsfasttemporalresolutionisthereverberation.Atransientsoundproducesacascadeofreectionswhosefrequencyofincidenceuponalistenergrowswiththesquareoftime;therateofarrivalofthesereectionsdN/dtistheroomvolume)approachesonceeverysafteronesecondfora2500mroom[2].Henceanaccuracyofreproductioninthemicrosecondrangeisnecessarytopreservetheoriginalacousticenvironmentsreverberation.Thepresentexperimentalresultthuspro-videsaconcretebasisfortheanecdotalclaimsbyau-diophilesofsensitivitytoveryshorttime-domainerrors(suchasaninsuciencyofsomecommonlyuseddigitalsamplingrates)asdiscussedintheIntroductionsection.Whiletheneurophysiologicalbasisthatunderliestheob-servedfasttemporaldiscriminabilityisnotofprimaryinterestforsoundreproduction,thepresentresultnever-thelessdoesshedsomelightonthisissue.Thestartingpointofanyhearingsensationinvolvesexcitationoftheinnerhaircellsinthecochlea.Threedeterminantscanchangetheperceptofasound.Oneisachangeinthestimulusfrequency/ies,whichchangestheCFsandloca-tionsalongthebasilarmembranewhereIHCsaremax-imallyexcited.Anotherisachangeinloudness,whichchangesthedegreeofexcitationandthewidthofthe cleusdetectcoincidentringofauditorynerveberswithtemporalprecision,J.Neurosci.,3138 3153(1995).[10]M.J.FerragamoandD.Oertel,Shapingofsynapticresponsesandactionpotentialsinoctopuscells,Assoc.Res.Otolaryngol.,96(1998).[11]D.H.Johnson,Theresponseofsingleauditory-nervebersinthecattosignletones:synchronyandaveragedischargerate,Ph.D.thesis,DepartmentofElectricalEngineering,MIT,Cambridge,MA(1974).[12]S.A.Shamma,N.Shen,G.Preetham,Stereausis:Bin-auralprocessingwihtoutneuraldelays,J.Acoust.Soc.,989 1006(1989).[13]I.Pollack,Submicrosecondauditoryjitterdiscrimina-tionthresholds,J.Acoust.Soc.Am.,1059 1059[14]I.Pollack,Spectralbasisofauditoryjitterdiscrimina-tion,J.Acoust.Soc.Am.,555(1971).[15]B.H.Deatherage,L.A.Jeress,andH.C.Blodgett,Anoteontheaudibilityofintenseultrasound,J.Acoust.Soc.Am.,582(1954).[16]F.J.Corso,Boneconductionthresholdsforsonicandultrasonicfrequencies,J.Acoust.Soc.Am.,1738 1743(1963).[17]M.L.Lenhardt,R.Skellett,P.Wang,andA.M.Clarke,Humanultrasonicspeechperception,Science,82 85(1991).[18]M.L.Lenhardt,Humanultrasonichearing,Hearing,50 52(1998).[19]S.Fujiokaetal.,BoneConductionHearingforUltra-sound,Trans.Tech.Com.Physio.Acoust.Soc.Japan,H-97-4(1997).[20]H.E.vonGierke,Subharmonicsgeneratedinhumanandanimalearsbyintensesound,J.Acoust.Soc.Am.,675(1950).[21]K.Ashihara,K.Kurukata,T.Mizunami,andK.Mat-sushita,Hearingthresholdforpuretonesabove20kHz,Acoust.Sci.&Tech.,12 19(2006).[22]T.Oohashi,E.Nishina,N.Kawai,Y.Fuwamoto,andH.Imai,High-frequencysoundabovetheaudiblerangeaectsbrainelectricactivityandsoundperception,J.AudioEng.Soc.(Abstracts),1010(1991).[23]S.Yoshikawa,S.Noge,M.Ohsu,S.Toyama,H.Yana-gawa,T.Yamamoto,Sound-qualityevaluationof96-kHzsamplingdigitalaudio,J.AudioEng.Soc.(Ab-,1095(1995).[24]M.JShailerandB.C.J.Moore,GapDetectionandtheAuditoryFilter:PhaseEectsUsingSinusoidalStimuli,J.Acoust.Soc.Am.,1110 1117(1987).[25]C.Formby,M.Gerber,L.Sherlock,andL.Magder,Evidenceforanacross-frequency,between-channelpro-cessinasymptoticmonauraltemporalgapdetection,J.Acoust.Soc.Am.,3554 3560(1998).[26]B.C.J.Moore,AnIntroductiontothePsychologyofHearing,5thedition,AcademicPress(2003).[27]R.PlompRateofdecayofauditorysensation,J.Acoust.Soc.Am.,277 282(1964).[28]M.J.PennerDetectionoftemporalgapsinnoiseasameasureofthedecayofauditorysensation,J.Acoust.Soc.Am.,552 557(1977).[29]D.A.Eddins,J.W.Hall,andJ.H.Grose,Detectionoftemporalgapsasafunctionoffrequencyregionandabsolutebandwidth,J.Acoust.Soc.Am.,1069 1077[30]D.P.Allen,T.M.Virag,andJ.R.Ison,Humansde-tectgapsinbroadbandnoiseaccordingtoeectivegapdurationwithoutadditionalcuesfromabruptenvelopechanges,J.Acoust.Soc.Am.,2967 2974(2002).[31]B.Leshowitz,Measurementofthetwo-clickthreshold,J.Acoust.Soc.Am.,462 466(1971).[32]D.Ronken,Monauraldetectionofaphasedierencebe-tweenclicks,J.Acoust.Soc.Am.,1091 1099(1970).[33]G.B.Henning,andH.Gaskell,Monauralphasesensi-tivitywithRonkensparadigm,J.Acoust.Soc.Am.1669 1673(1981).[34]K.Krumbholz,R.D.Patterson,A.Bobbe,andH.Falstl,Microsecondtemporalresolutioninmonauralhearingwithoutspectralcues?,J.Acoust.Soc.Am.,2790 2800(2003).[35]K.Ashihara,andS.Kiryu,Inuenceofexpandedfre-quencybandofsignalsonnon-linearcharacteristicsofloudspeakers,NipponOnkyoGakkaiShi(J.Acoust.Soc.Jap.),549 555(2000).[36]InternationalStandardsOrganizationminimumaudibleeld(MAF)standard:ISO389-7(1996).[37]K.Kurukata,T.Mizunami,K.Matsushita,andK.Ashihara,Statisticaldistributionofnormalhearingthresholdsunderfree-eldlisteningconditions,Acoust.Sci.&Tech.,440 446(2005).[38]P.Rogowski,A.Rakowski,andA.Jaroszewski,Spe-cicHearingLossinYoungPercussionandBrassWindPlayersduetoMusicNoiseExposures,The8thInter-nationalCongressonSoundandVibration,HongKong,China,2 6July(2001).[39]J.Boyk,Thereslifeabove20kilohertz!Asur-veyofmusicalinstrumentspectrato102.4kHz,http://www.cco.caltech.edu/boyk/spectra/spectra.htm.Copyright1992,1997JamesBoyk,MusicLab,Califor-niaInstituteofTechnology.[40]W.Jesteadt,C.C.Wier,andD.M.Green,Intensitydiscriminationasafunctionoffrequencyandsensationlevel,J.Acoust.Soc.Am.,169 177(1977).[41]R.PlompandH.J.M.Steeneken,Theeectofphaseonthetimbreofcomplextones,J.Acoust.Soc.Am.409 421(1969).[42]M.N.Kunchur,Temporalresolutionofhear-ingprobedbybandwidthrestriction,ActaAcusticaunitedwithAcustica,594-603(2008).(Preprintcanbedownloadedfromhttp://www.physics.sc.edu/kunchur/temporal.pdf)