Page 1
2/15/18
1
Spa$al Audio for VR: An Overview
What is spa$al audio?
• alsoreferredtoas3Daudioor360audio• sonicexperiencewhere
• theaudiochangeswiththemovementoftheviewer’shead
• producedbystereospeakers,surround-soundspeakers,speaker-arrays,orheadphones.
• Spa6almusicismusiccomposedtointenConallyexploitsoundlocalizaCon
• inusesinceprehistoricCmesintheformofanCphon• inusesincearound1928as‘Raumusik’or“spacemusic”fromGermany
Page 2
2/15/18
2
Spa$aliza$on:
• theprojecConandlocalizaConofsoundsourceinaspace,• Physical• simulated
• anditsspaCalmovementinspace.
• technicallyknownasspaCaldomainconvoluConofsoundwavesusinghead-relatedtransferfunc6ons(HRTF).
Crea$ng posi$onal sound
• Amplitude• (ormore)
• SynchronisaCon• Audiodelays
• Frequency• Head-RelatedTransferFuncCon(HRTF)
Page 3
2/15/18
3
Amplitude• GenerateaudiofromposiConsources• Calculateamplitudefromdistance• Includedampingfactors
• AircondiCons• Snow• DirecConaleffectoftheears
Synchronisa$on
• Earsareverypreciseinstruments• Verygoodathearingwhensomethinghappensa]ersomethingelse
• Soundtravelsslowly(c340m/secinair):differentdistancetoeachear
• UsethistohelpdefinedirecCon• DifferenceinamplitudegivesonlyveryapproximatedirecConinformaCon
Page 4
2/15/18
4
Speed effect
• 30cenCmetres=0.0008seconds• Humancanhear≤ 700µS
What is 3D sound?
• AbletoposiConsoundsallaroundalistener.• Soundscreatedbyloudspeakers/headphones:perceivedascomingfromarbitrarypointsinspace.
• ConvenConalstereosystemsgenerallycannotposiConsoundstoside,rear,above,below
• Somecommercialproductsclaim3Dcapability-e.gstereomulCmediasystemsmarketedashaving“3Dtechnology”.Butusuallyuntrue.
Page 5
2/15/18
5
3D posi$onal sound
• Humanshavestereoears• Twosoundpulseimpacts
• Onedifferenceinamplitude• OnedifferenceinCmeofarrival
• Howisitthatahumancanresolvesoundin3D?• Shouldonlybepossiblein2D?
Frequency
• FrequencyresponsesoftheearschangeindifferentdirecCons• Roleofpinnae• Youhearadifferentfrequencyfilteringineachear• Usethatdatatoworkout3DposiConinformaCon
Page 6
2/15/18
6
Head-Related Transfer Func$on
• UnconscioususeofCmedelay,amplitudedifference,andtonalinformaConateacheartodeterminethelocaConofthesound.
• Knownassoundlocalisa+oncues.• SoundlocalisaConbyhumanlistenershasbeenstudiedextensively.
• TransformaConofsoundfromapointinspacetotheearcanalcanbemeasuredaccurately
• Head-RelatedTransferFuncCons(HRTFs).• MeasurementsareusuallymadebyinserCngminiaturemicrophonesintoearcanalsofahumansubjectoramanikin.
HRTFs
• HRTFsare3D• Dependonearshape(Pinnae)andresonantqualiCesofthehead!• AllowsposiConalsoundtobe3D
• ComputaConallydifficult• Originallydoneinspecialhardware(Convolvotron)• Cannowbedoneinreal-CmeusingDSP
Page 7
2/15/18
7
HRTFs
• FirstseriesofHRTFmeasurementexperimentsin1994byBillGardnerandKeithMarCn,MachineListeningGroupatMITMediaLab.
• Datafromtheseexperimentsmadeavailableforfreeontheweb.
• PictureshowsGardnerandMarCnwithdummyusedforexperiment-calledaKEMARdummy.
• A measurement signal is played by a loudspeaker and recorded by the microphones in the dummy head.
HRTFs
• Recordedsignalsprocessedbycomputer,derivestwoHRTFs(le]andrightears)correspondingtosoundsourcelocaCon.
• HRTFtypicallyconsistsofseveralhundrednumbers• describesCmedelay,amplitude,andtonaltransformaConforparCcularsoundsourcelocaContole]andrightearsofthesubject.
• MeasurementprocedurerepeatedformanylocaConsofsoundsourcerelaCvetohead
• databaseofhundredsofHRTFsdescribingsoundtransformaConcharacterisCcsofaparCcularhead.
Page 8
2/15/18
8
HRTFs
• Mimicprocessofnaturalhearing• reproducingsoundlocalisaConcuesattheearsoflistener.
• UsepairofmeasuredHRTFsasspecificaConforapairofdigitalaudiofilters.
• Soundsignalprocessedbydigitalfiltersandlistenedtooverheadphones• ReproducessoundlocalisaConcuesforeachear• listenershouldperceivesoundatthelocaConspecifiedbytheHRTFs.
• Thisprocessiscalledbinauralsynthesis(binauralsignalsaredefinedasthesignalsattheearsofalistener).
HRTFs
• The process involved in generating true 3D audio using HRTFs:
Page 9
2/15/18
9
Fourmajordivisionsofauditorysystem–Func6on
I. Outer ear: (1) Pinna (Binaural cue to sound source loca$on)
t Left
t Right
*Differentdistancesfromsourcetoeachear=>differentarrival6mes(Interaural6me-difference)anddifferentsoundlevel(interaurallevel-difference)
Rightear
LeNear
Sound
Page 10
2/15/18
10
I. Outer ear: (1) Pinna (Spectral cue to sound source loca$on)
Out
er e
ar g
ain
-15° 0°
+15°
ThespectralfeatureofsoundischangeddependingonthesoundelevaCon=>HeadRelatedTransferFunc6on(HRTF)
Basics of Object Based Audio WhatisanObjectBasedAudio?
Audiowhichisgeneratedby
• aStaConaryormovingobjectOR
• aclassofobjectsthatareclubbedtogetherasacollecCvesourceofsound
SomeExamplesofAudioObjectsare:
• AStagearCste
• Chorus
• CheeringCrowdatCricketground
• Oceanwaves,windblowing
• Birdschirping
• AnAirplaneorAHelicopter
• AmovingTrain
• Abulletfired
• Amonologueoradialogue
Page 11
2/15/18
11
Examples of Object-Based Audio AudioSceneorSoundFieldgeneratedbymixing(notjustadding)audiosignalsfrommulCpleObjects.Followingareafewexamples.
• Watchingafootballmatchinastadiumwithhomecrowd.(SportsTVChannel)3objects:HomeCrowdasaringobject,Commentatorasapointobject,Players-Umpire conversaConsasanotherobject.
• ParCcipaCngasaplayerinafieldgame.(ComputerGames)4+objects:HomeCrowdasaringobjectaroundyou,Commentatorasapointobject, Player’sownvoiceresponsesasapointobject, OtherPlayers-UmpireasmulCplemovingobjects.
• BeingapartofScuba-diverteamsearchinganunderwatertreasure.(VR)
• ListeningtoaconversaConbetweendifferentactors&backgroundsinamoviescene.(cinema)
• ArendingamusicconcertorasimpleJazzperformance(concert)
ChannelBasedImmersiveAudio ObjectBasedAudio
ContentCreaCon
• Eachsignaltrackisassociatedwithaspecificspeakerfeed&setupatlistenerend.
• ContentiscreatedforaspecificListenerEnvironmentorsetup.(mobile,home,ortheater)
• AudioObjectbasedsignaltracksareindependentofspeaker-setup.
• =>ContentcreatedisindependentListenerEnvironmentorsetup.(mobile,home,ortheater)
PlaybackatListenerEnd
• AtListenerend,thecontents(channels)aremappedontouserspeakersetup
• NeedtousePredefinedchannel-mappingtoheadphones,stereospeaker,2.1,5.1,11.1etc.
• AtListenerend,theobjectsaremappedontouserspeakersetup
• ObjectsbasedonposiConsandmovementsaremappedontheflytothespeaker-setup.
Channel Based Audio vs Object Based Audio
Page 12
2/15/18
12
Channel Based Audio vs Object Based Audio
ChannelBasedImmersiveAudio ObjectBasedAudio
ContentCreaCon
• Withinputsastherecordedcontentsortracks,eachChanneltrackiscarefullydesignedandcreatedattherecordingstudios.ORatthegamingdeveloperstudiosforcreaCnggoodimmersiveeffects.
• AudioObjectscanbesimplyidenCfiedencodedasseparatetracks.
• Associatedmeta-datashouldbecarefullydesignedtocaptureshape,movement,appearance/disappearanceoftheobjectsassumingthelisteneratthecenter.
PlaybackatListenerEnd
• Ifthecontent-targetspeaker==userspeakersetup,thensimple-mappingandplayback.
• Elseusesomegoodpre-definedmapsanddelaysforrearspeakerstocreatethecontent.
• Objectsaredecodedtocreateaudiosignals.• Frame-by-Frame,posiConsof“acCveobjects”
aremappedontouserspeakersinformofgainsanddelaysfortheseobjects.Mixandplayback.
ChannelBasedImmersiveAudio ObjectBasedAudio
ContentCreaCon
• CreaConisacomplexcarefulprocess.• Encodingstepsandprocedureiscomplexand
henceisdonebyskilledwelltrainedsounddesigners.
• CreaConandencodingobject-audioisarelaCvelysimplerprocessandcanbedonewithoutmuchpre-thinkingofuser-setups&environment.
• Audioobjectmeta-dataneedstobecarefullyassociatedwithit.
PlaybackatListenerEnd
• DecodersareRenderersarefairlysimple.
• Decodersaresimple(assimpleaschannelbasedAudio).
• HowevertheRenderersaremuchmorecomplex.
• RenderersneedtomaptheseobjectswithitsposiConstospeakersonaframe-by-framebasis.
Channel Based Audio vs Object Based Audio
Page 13
2/15/18
13
A summary – basic of Object-Based Audio
Enc-Audio-ObjFrame1
Enc-Audio-ObjFrame2
Enc-Audio-ObjFrameN
MetaData-ObjFrame1
MetaData-ObjFrame2
MetaData-ObjFrameN
ObjectAudioTrack
TypicalObject-AudioEncodedStreamcontains8to16encodedaudioobject-tracks.Andeachaudio-objecttrackhastwoparts
Meta-Data:
Stream-level:• MaxNumberofAudioObjectspresentinthescene.
Frame-level:• Shapeoftheobject,DatarelatedtoposiCon,speedof
theobject,Appearance/Disappearanceofobject
CompressedPCMAudio-Data:
StandardDDorAACencodedaudiosignalsassociatedwithaspecificobject.
Object-Based Audio Stream Decoding & Rendering
• Decoding:Thebasicaudiofromobjectisencodedusingstandardlegacyencoders.Therefore,decodingusesstandardmp3,aac,dolby-digitaldecodingtoprovidebasicaudioPCMfortheobject.
• Renderers:ChallengesareinRenderingthedecodedobject-basedaudioPCMcontents&useobject’sshape/moConmeta-datatocreate–
• Animmersiveaudioexperienceonheadphones.(VR,Gaming,smartphones,andtablets)
• Animmersiveaudioexperienceonourmul6-speakerlayoutsathomesortheaters
Page 14
2/15/18
14
Object-Based Audio Renderer on Headphones
Diagram of spherical coordinate system(WightmanandKistler, Univ Winsconsin) – 1989 [1]
Earimpulse&frequencyresponsefororientaConshowninpictureonle](90ºazimuth,0º-elevaCon)fromWKSDOset.[2,3](Note:5to9msecduraConimpulseresponsewidth@16kHzsamplerate)
HRTFModel(HeadRelatedTransferFuncCon)
Object-Based Audio Rendering on Headphones
DecodedaudioobjectPCMdata
HRTF_L/R–Filterpoolfordiffvaluesof
φi & θi
DecodedaudioobjectMetadata
(dist,azi,ele)
Gain/delayModule
Distance->Gaindelaymapping
φ & θ
Interpolate&ComputeHRTF_Le]&
HRTF_Rightfromfilter-pool
HRTF_Le]
HRTF_Right
FIRFilterPair
R
MIXER
Page 15
2/15/18
15
Object-Based Audio Rendering on Headphones Signal
Processing
Challenges
PresetFIRFilterpool
DecodedaudioobjectPCMdata
DecodedaudioobjectMetadata(dist,azi,ele)
GaindelayModule
Distance->Gain,delaymapping
φ & θ
ComputeHRTF_Le]&HRTF_Right
HRTF_Le]
HRTF_Right
FIRFilterPair
R
MIXER
• TheparametersR, φ & θchangeeveryframe(20-30msec)⇒ filtercoeffs,gainchangeeveryframe⇒ Thismaycauseglitches,distorConsintheoutputs.⇒ NeedfortechniquestoadapCvely&smoothlychangethosecoefficients
• TherearemulCpleobjects&someappearanddisappeara]erafewframes.⇒ Needforontheflyobjectpcm+associatedmeta-datamemory
allocaCon,updateanddestrucCon⇒ Needforfade-in/fade-out/muteofoutputPCMsamples⇒ Needforawell-designedmulC-portPCMmixingmodule
• SomeObjectsmoveveryrapidly⇒ “R”changesw.r.t.Cme=>thespeedoftheobjectissubstanCalcausing
Dopplereffectonaudiosignal(e.g.afast-trainpassingby)⇒ Needforpitchshi]ing(variable-delay)moduletobeintroducedontopof
gainapplicaConModule.Oversampling&InterpolaConwouldberequired.
• VR/ComputerGamesrelated:Head/JoysCckmovementschanges“R,φ & θ”⇒ AnaddiConalHead-TrackingorJoysCckmovementsmodulewhichfeeds
userorientaConparameters“Ru,φu & θu”⇒ AddiConalmoduletoPerform3-DgeometrycomputaConstoderivefinal
objectposiConparametersfromtheabove2sets“R,φ & θ”
Object-Based Audio Rendering on Immersive Speaker-Layouts
DTS/DD+7.1SpeakerLayoutDolbyATMOS11.1SpeakerLayouts
ExamplesofImmersiveSpeakerLayouts
Page 16
2/15/18
16
Object-Based Audio Renderer on Immersive Speaker-Layouts
DTS-X7.2.4SpeakerLayout DTSNeo:X11.1SpeakerLayout
ExamplesofImmersiveSpeakerLayouts
Object-Based Audio Renderer on Immersive Speaker-Layouts
Ambisonics13.1SpeakerLayoutAuro-3D11.1SpeakerLayout
ExamplesofImmersiveSpeakerLayouts
Page 17
2/15/18
17
Object-Based Audio Renderer on Immersive Speaker-Layouts
VBAP(VectorBasedAmplitudePanning):• Alargearrayof“Virtual”SpeakerPosiConsareassumedtosurroundthelistener.Audio-
ObjectsandtheirmoCons/posiConsw.r.t.thelisteneraremappedonalargersetof“Virtual”SpeakerPosiCons.
• AudiosignalsforeachobjectismappedonthisvirtualspeakerposiConsusingVBAP
method
• Theaudioassociatedwithvirtualspeakersisthenmappedtostandarduserspeakerlayoutsusingpre-defineddown-mixingmatrices&setofdelays.
TwoMainTechniques.
• VBAP–VectorbasedamplitudePanning:MappingobjectaudiotoVirtualSpeakerArray
• HOA–HigherOrderAmbisonics:CreaCngdesired“Sound-Field”atlisteners’si}ngposiCon
HigherOrderAmbisonics [Gerzon1970]
• Createsasoundfieldgeneratedbyaudio-object(s)whenitgetscapturedbydirec6onalmicrophoneslocatedatthelistener’sposiCon
HOA based object rendering on Immersive Speaker-Layouts
FirstOrderAmbisonicsfields
SecondOrderAmbisonicsfields
AnAmbisonicMicrophone
• HOAchannelsareencodedandthesechannelsaredecodedandthenmappedontoanystandard“userspeakerlayouts”from5.1,or7.2.4or13.1”.Thesemappingsareeasyandlesscomplex.
• HOAtechniquemakesiteasytomodifythesound-fieldfordifferentuser(listener)orientaCons(requiredmainlyinVR&ComputerGaming)
Page 18
2/15/18
18
Ambisonics 3D (1st order)
Reproduction occurs over an array of 8-24 loudspeakers, through an Ambisonics decoder
Original Room
Sound Source
SoundField Microphone
B - format 4 - channels signal
(WXYZ)
Ambisonics decoder
Speaker array in the reproduction room
3D extension of the pressure-velocity measurements • The Soundfield microphone allows for simultaneous measurements of the omnidirectional
pressure and of the three cartesian components of particle velocity (figure-of-8 patterns)
Page 19
2/15/18
19
A-format microphone arrays • Today several alternatives to Soundfield microphones do exists. All of them are providing “raw” signals
from the 4 capsules, and the conversion from these signals (A-format) to the standard Ambisonic signals (B-format) is performed digitally by means of software running on the computer
Ambisonics decoding
• Each speaker fed is just obtained as a weighted sum of the 4 B-format signals (WXYZ)
• The weighting factors only depend on the position of each loudspeaker
• It is possible to add a small FIR filter for matching perfectly the gain and phase of all loudspeakers
r
y
z
x
α
β
γ
Page 20
2/15/18
20
Ambisonics decoding advantages: n Tridimensional n Good later sound perception n Good bass response n Wide “sweet spot”, no colouring outside
it
disadvantages: n Not isotropic n Advanced decoding required (Y gets
more weight than X and Z, and this must be compensated for)
Bi-square Ambisonics array
Bi-square Ambisonics array
8 Turbosound Impact 50 loudspeakers: n Light, easily fixed and oriented n Good frequency response n Very little distortion
Front-Right
Down-Right
Up-Right
Page 21
2/15/18
21
What is Ambisonics? • Ambisonicsisamethodforrecording,mixingandplayingbackthree-dimensional360-degreeaudio.Itwasinventedinthe1970sbutwasnevercommerciallyadoptedunClrecentlywiththedevelopmentoftheVRindustrywhichrequires360°audiosoluCons.
• ThebasicapproachofAmbisonicsistotreatanaudiosceneasafull360-degreesphereofsoundcomingfromdifferentdirecConsaroundacenterpoint.
• Thecenterpointiswherethemicrophoneisplacedwhilerecording,orwherethelistener’s‘sweetspot’islocatedwhileplayingback.
Ambisonics B-format
• ThemostpopularAmbisonicsformattoday,widelyusedinVRand360video,isa4-channelformatcalledAmbisonicsB-format
• usesasfewasfourchannelstoreproduceacompletesphereofsound.
Page 22
2/15/18
22
Ambisonics vs. Surround
• TradiConalsurroundtechnologiesaremoreimmersivethansimpletwo-channelstereo,buttheprinciplebehindthemisthesame:
• Theyallcreateanaudioimagebysendingaudiotoaspecific,pre-determinedarrayofspeakers.
• Stereosendsaudiototwospeakers;5.1surroundtosix;7.1toeight;andsoon.
• Bycontrast,AmbisonicsdoesnotsendaudiosignalstoanyparCcularnumberofspeakers;itis“speaker-agnosCc.”Instead,Ambisonicscanbedecodedtoanyspeakerarray.Ambisonicaudiorepresentsafull,uninterruptedsphereofsound,withoutbeingrestrictedbythelimitaConsofanyspecificplaybacksystem.
Ambisonics as standard in 360 video and VR: • TradiConalsurroundformatscanprovidegoodimagingwhenstaCc;butasthesoundfieldrotates,thesoundtendsto‘jump’fromonespeakertoanother.
• Ambisonicscancreateasmooth,stableandconCnuoussphereofsound,evenwhentheaudioscenerotates(as,forexample,whenagamerwearingaVRheadsetmovesherheadaround).ThisisbecauseAmbisonicsisnotpre-limitedtoanyparCcularspeakerarray,
• TradiConalsurroundspeakersystemsareusually‘front-biased’:informaConfromthesideorrearspeakersisnotasfocusedasthesoundfromthefront.Bycontrast,Ambisonicsisdesignedtospreadthesoundevenlythroughoutthethree-dimensionalsphere.
• Finally,whereastradiConalsurroundsystemshavevariousdifficulCesrepresenCngsoundbeyondthehorizontaldimension,Ambisonicsisdesignedtodeliverafullspherecompletewitheleva'on,wheresoundsareeasilyrepresentedascomingfromaboveandbelowaswellasinfrontorbehindtheuser.
Page 23
2/15/18
23
First-order Ambisonics B-format
• Thefourchannelsinfirst-orderB-formatarecalledW,X,YandZ.OnesimplifiedandnotenCrelyaccuratewaytodescribethesefourchannelsistosaythateachrepresentsadifferentdirecConalityinthe360-degreesphere:center,le]-right,front-back,andup-down.
• AmoreaccurateexplanaConisthateachofthesefourchannelsrepresents,inmathemaCcallanguage,adifferentsphericalharmoniccomponent–or,inlanguagemorefamiliartoaudioengineers,adifferentmicrophonepolarpacernpoin6nginaspecificdirec6on,withthefourbeingcoincident(thatis,conjoinedatthecenterpointofthesphere).
First-order Ambisonics B-format
• Wisanomni-direcConalpolarparern,containingallsoundsinthesphere,comingfromalldirecConsatequalgainandphase.
• Xisafigure-8bi-direcConalpolarparernpoinCngforward.• Yisafigure-8bi-direcConalpolarparernpoinCngtothele].• Zisafigure-8bi-direcConalpolarparernpoinCngup.
Page 24
2/15/18
24
X, Y, and Z channels:
• Afigure-8microphonehasaposiCvesideandanegaCve(inversephase)side.WhiletheXchannel’sfigure-8polarparernpointsforwards,itsnegaCvesidepointsbackwards.TheresulCngaudiosignalontheXchannelcontainsallthesoundthatisinthefrontofthespherewithposiCvephase,andallthesoundsfromthebackofthespherewithnegaCvephase.
• ThesamegoesfortheYandZchannels:TheYchannelspickupthele]sideofthespherewithposiCvephaseandtherightsidewithnegaCvephase.TheZchannelpicksupthetopsideofthespherewithposiCvephaseandtheboromwithnegaCvephase.Thisway,bymeansofdifferenCalgainandphaserelaCons,thefourchannelscombinedrepresenttheenCrethree-dimensional,360-degreesphereofsound.
AmbiX vs. FuMa
• twoconvenConswithintheAmbisonicsB-formatstandard:AmbiXandFuMa.Theyarequitesimilar,butnotinterchangeable:theydifferbythesequenceinwhichthefourchannelsarearranged,withAmbiX,forexample,arrangedWYZXinsteadofWXYZ.
Page 25
2/15/18
25
First-order to sixth-order Ambisonics • The4-channelformatisonlyasimple,first-orderformofB-format,whichiswhatmostAmbisonicsmicrophonesandplaybackpla�ormssupporttoday.
• Higher-orderB-formataudiocanprovideevenhigherspaCalresoluCons,withmorechannelsprovidingmoredifferentpolarparerns.
• Second-orderAmbisonicsuses9channels,• Third-orderAmbisonicsuses16channels,• Sixth-orderAmbisonicsuses49channels.
Playing back Ambisonics (DxARTS, Rai` Hall)
Room117uses24full-rangespeakersand4subwoofersforfullheightspaCalsoundreproducCon.ThereisarouCngdecodingsystemtocareofpatching,ambisonicdecoding,speakerbalancingandroomcorrecCon,andthecrossoversfordistribuCngsoundtothesubs.YoucansendaB-formatsignaltovariousdecodersoraddresseachspeakerindividually,dependingonse}ngs.
Page 26
2/15/18
26
Decoding Ambisonics
• YoucanplaybackAmbisonicsonalmostanyspeakerarray,recreaCngthesphericalsoundfieldatthelisteningspot.Buttodothat,youneedtodecodethefourB-formatchannelsforthespecificspeakerarray.
• AllfourB-formatchannelsaresummedtoeachspeakerfeed.Eachofthefourchannelsissummedwithdifferentgainandphase,dependingonthedirecConofthespeaker.
• Someofthesourcesinthemixaresummedin-phasewhileothersaresummedout-of-phaseateachspecificspeaker.
• TheresultisthatsourcesalignedwiththedirecConofthespeakerarelouder,whilethosenotalignedinthedirecConofthespeakerarelowerorcancelout.
Page 27
2/15/18
27
Ambisonics on headphones
• SpaCalsoundonheadphoneismadepossiblebybinauralaudiotechnologies.Inessence,abinauralprocessorreceivesanaudioinputandadirecConinwhichtoposiConit.theprocessoraddsauditorycuestothesignal,sothatwhenplayedbackonheadphonesitisexperiencedatthesetvirtualposiCon.
• ThemostcommonwaytoprocessAmbisonicsforbinauralspaCalplaybackonheadphonesistodecodetheAmbisonicschannelsforacertainspeakerarray–andsendthefeedstoabinauralprocessorwhichvirtuallyposiConsthematthedirecConthattheactualspeakerwouldhavebeen.
• Theresultisthattheimmersivesphericalsoundfieldisexperiencedbythelistenerwhenmonitoringonheadphones.
Ambisonics on headphones
Page 28
2/15/18
28
BinAural Rendering : Immersive Speakers -> Headphones. Dependinguponthefixedφ & θ anglesofFront,Rear&OverheadSpeakersw.r.t.Le]andRightearofthelistener
HRTF_Le]andHRTF_rightareapplied&
mixedasbelow.
HRTF_Le]Ear_Lf
HRTF_RightEar_Lf
Le]FrontSpeakerSignalLf
HRTF_Le]Ear_Rf
HRTF_RightEar_Rf
RightFrontSpeakerSignalRf
HRTF_Le]Ear_Lsr
HRTF_RightEar_Lsr
Le]RearSpeakerSignalLsr
HRTF_Le]Ear_Rsr
HRTF_RightEar_Rsr
RightRearSpeakerSignalRsr
+
HRTF_Le]Ear_Oh
HRTF_RightEar_Oh
OverheadSpeakerSignalOh
+
Content plaborms that support spa$al audio
• YouTube• Facebook360SpaCalWorkstaCon• GoogleVRResonanceAudio• SamsungVR• Others
Page 29
2/15/18
29
Facebook 360 Spa$al Worksta$on
• TheFacebook360SpaCalWorkstaConisaso]waresuitefordesigningspaCalaudiofor360videoandcinemaCcVR.ItincludespluginsforpopularaudioworkstaCons,aCmesynchronized360videoplayeranduCliCestohelpdesignandpublishspaCalaudioinavarietyofformats.AudioproducedwiththetoolscanbeexperiencedonFacebookNewsFeedonAndroidandiOSdevices,ChromefordesktopandtheSamsungGearVRheadsetthroughheadphones.
Google VR Resonance Audio ResonanceAudiogoesbeyondbasic3DspaCalizaCon,providingpowerfultoolsforaccuratelymodelingcomplexsoundenvironments.• TheSDKenables:
• SoundsourcedirecCvitycustomizaCon• Near-fieldeffects• Soundsourcespread• Geometry-basedreverb• Occlusions• RecordingofAmbisonicaudiofiles
• 3rdorderAmbisonicsformatssupported
Page 30
2/15/18
30
The end…