AND

PRESENT

Acoustical Measurement
StandardsforStereo
ListeningRooms

NyalMellor,AcousticFrontiersLLC&JeffHedback,HdAcoustics
2011

Thispaperpresentsinroomacousticalmeasurementstandardsaswellasquantifiersofthe
environmentitselfforthepurposeofhighperformance2channelaudioreproduction.Contextand
examplesareofferedinvariousmetricsrangingfromnoise,timerelatedelements,frequencyresponse,
roomsizeandroomconstruction.Thevariousindividualsectionsarecoveredeachwithintheirown
realmandinhowtheyinterrelatetowardtheresultinglisteningexperience.

Introduction
Muchhasbeenwrittenaboutacousticalstandardsforhometheaterandprofessionalstudiocontrol
rooms.Verylittle,however,hasbeencommittedtopaperaboutacousticalstandardsforhighend
stereoortwochannelmusiclisteningrooms.Theverypracticeoflocatingonlytwospeakersanda
singlelisteninglocationinaroomleadstoconsequencesthataretoooftenignoredandcertainlypoorly
understood.InthiscoauthoredwhitepaperNyalMellorofAcousticFrontiersandJeffHedbackof
HdAcousticsaimtoexplainabasicsetofacousticalstandardsthatcanbeusedintheevaluationof
existingroomsorthedesignandpostconstructionverificationofanewroom.
Whilstsomeofthestandardsforhometheaterandstudioscanbeseentohaveapplicabilityfor
residentialstereolisteningroomsmanydonot.Instudios,forexample,thehotseatisalmostalways
nearfieldduetothelocationofspeakersontopofamixingconsole,puttingtheminalocationwhere
theoverallsoundbalanceisdominatedbythedirectenergyfromthespeakers.Thisisaverydifferent
acousticalenvironmentthaninahomewherereflectedsoundsfromtheroomarestrongerthanthe
directsound.Theothertypicalstudioconfigurationisforthespeakerstobeflushorsoffitmounted,
whichsignificantlychangesthewayinwhichthespeakerinteractswiththeroom.Inahometheaterthe
prioritiesaredifferent:whilstmaintainingspectralbalanceisanequalpriorityinbothsituationsdigital
roomcorrectionisoftenusedinatheater.Acousticaldesignthenfocusesonminimizingseattoseat
variabilityinfrequencyresponsethroughcarefulstudyofthenumberandlocationofmultiple
subwoofers.Thispracticeallowssubsequentapplicationofequalization.Moreover,ahometheateralso
hassurroundspeakerswhichmakethegenerationofenvelopmentandthefeelingofbeingimmersedin
athreedimensionalsoundfieldsignificantlyeasiertoachievethaninatwochannelenvironment.Itis
thereforeimperativethatwedonotjustblindlyexpectwellknownstudioandhometheaterdesign
conceptstoportstraightintoatwochannelenvironment.Wemustdevelopourownsetofacoustical
standardsthattakeintoaccounttheuniqueaspectsoftwochannelreproductioninthehome
environment.
Withtherecentavailabilityoffreeorlowcostprosumeracousticalmeasurementsystemssuchas
RoomEQWizard,XTZRoomAnalyzerandPartsExpressOmnimicdedicatedenthusiastscannowtakea
setofmeasurementsthatdescribetheacousticalperformanceoftheirexistingroom.Thesesystems
makeiteasytotakemeasurementsbuttheydonotprovideguidanceonwhatagoodorbad
measurementlookslike.Thislackofguidancefromthesoftware,andthelackofanyformalstandards
fromtheacousticalpractitionercommunitythatarespecifictothechallengesoftwochannel
reproductionmakethiswhitepapersorelyneeded.
Thispaperputsforwardasetoftargetsforwhattheacousticalmeasurementsofahighperformance
twochannelmusiclisteningroomshouldlooklike.Wedescribewhatthekeymeasurementsare,what
anoptimalmeasurementwouldbeandarangeofacceptability.Oncearoomsmeasurementshave
beencomparedwithtargetstheendusercanthendecidewhethertogothepathofeducating
themselvesonacoustics,acoustictreatmentandacousticdesignbyreadingbooksandexperimenting
and/orhiretheservicesofanacousticaldesignertocorrecttheirroomsacousticaldeficiencies.

Page2of30

Summaryofacousticalstandards
Criteria

Measurement

Target

AcceptableRange*

A:NoiseControl

RC

RC20

RC30

B:ReflectedSound

EnergyTimeCurve,040ms

L&Rspeakersvisuallyidentical

~10dBreductioninenergylevel
by40ms
Cleardecreaseinenergyoverthe
40ms
Peakssmoothinpatternand
density
C:LowFrequencyDecay
Times

LowFrequencyResonance**

ReferencechartFigureC.1

ReferencechartFigure
C.1

D:MidrangeDecayTimes

T20,T30,T60

T60between0.2and0.5s

T20,T30deviation<25%across
onethirdoctavebands
E:MidrangeFrequency
Response

MidrangeFrequency
Response

Lspeakerwithin+/3dBat1/3rd
octave

Rspeakerwithin+/3dBat1/3rd
octave
Nodeviationgreaterthan3dB
betweenL/Rspeakers
F:LowFrequency
Response

LowFrequencyResponse**

G.RoomSizeand
Construction

Physicaldimensions

Within+/10dBat1/24thoctave

Within+/5dBat1/3rdoctave
Sizeofroom224sqftto475sqft

Volumeofroom1,750cuftto
4,750cuft
Dimensionslackcommondivisors
(2x)layersof5/8gypsumboard
Surfacesareconstructedusing
similarmethods
*Notallmeasurementshavealowerlevelperformancetargetdefined.
**Allmeasurementsshouldbecarriedoutforeachspeakerseparatelyexceptthelowfrequencymeasurementswhichshould
becarriedoutwithbothspeakersplayingtogether.Itisimportanttoensurethatthemeasurementlevelsareconsistentwhen
comparingleftandrightspeakersotherwiseerroneousconclusionsmaybedrawnfromexaminationofthedata.

Page3of30

A:NoiseControl(indicatedbyRoomCriterionRCrating)
Standard:NoisecriterionbelowRC30forexistingroomsorRC20forpurposebuiltrooms
Highnoiselevelscancauseauditorymaskingoflowleveldetailswithinrecordingsthatreduce
soundstagedepthandtimbralcolorby:

Obscuringthewayinwhichsounddecayswithinarecordingvenue.Eachrecordingvenuehasa
characteristicacousticsignaturethatcanbelostifnoiselevelsaretoohigh.Listeningtoclassical
music,itmightbedifficulttodifferentiateaperformanceintheDisneyHallrelativetothe
CarnegieHall.StandardsforconcerthallsandrecordingstudiosaretypicallyRC1520andone
couldarguethatareproductionenvironmentshouldhaveanequalorlowernoisefloorthanthe
recordingvenueiftheacousticsignatureofthatvenueistobereproducedfaithfully.

Obscuringthelowlevelharmonicsthatgiveeachmusicalinstrumentitscharacteristictimbre.
OnemightfinditdifficulttodifferentiateaBosendorferfromaSteinwaypiano.Thereisadirect
relationshipbetweenthenoisefloorofaspaceandtheabilitytoperceivethehighestlevelsof
audiomicrodetail.

ThenoiseinanexistingspaceshouldnotbeaboveRC30.Thiseliminatesmostdistractionsthat
otherwiseinterfereswiththelisteningexperience.Typicalresidentialspaceswillhavebackgroundnoise
levelsintheRC3545range.AratingaslowasRC25shouldbeachievablewithincommonductedHVAC
systemswithremediationinareasofinternalacousticalliningsandacousticallyratedgrilles/diffusors(as
justtwoexamplesofthetypesofchangesrequired).Itisimportanttofactorotheritemsliketheaudio
equipmentitselforcomfortelementssuchaswinechillersthatmaycontributetothebackgroundnoise
levelofthelisteningenvironment.FornewdedicatedroombuildswesuggestatargetofRC20.Meeting
RC20willrequirepayingattentiontonoiseisolationandusingspecializedconstructiontechniques.
TherearecertainlybenefitstoaratingofRC15andthispaperisinnowaydiscouragingthegreater
potentialofroomdynamicsandultrablacknoisefloor.Itisagreedforthepurposesofthispaperthat
theexponentialincreasestoroomconstructioncostsmakeestablishmentofRC15asastandard
unrealistic.

Page4of30

B:ReflectedSounds
Standards:
ETCoftheL&Rspeakersshould:

Bevisuallyidentical(withonlyminordeviations)from040ms

Bedownto10dBby40mstopreventbreakdownoftheprecedenceeffect

Clearlyshowadecreaseintheamplitudeofenergyover040ms.Thedecaypatternmay
ormaynotbecontinuous.

Showtheconsecutivepeaksofthehighestamplitudereflectionsviewedacrossthetime
axistoberelativelysmoothinpatternanddensity.

Thecriteriaaboveshouldbeconsideredinconjunctionwiththeotherstatedtargetsfor
reverberationtimeandL/Rfrequencyresponse.

BeforewegetstarteditisusefultounderstandwhatanETCisandhowitdiffersfromafrequency
responsechart.Frequencyresponse,asshownbyanFFTofanimpulseresponse,addsupallincident
soundwithintheimpulseresponse(IR)gatestoproducetheamplitudevs.frequencyplot.AnETCisa
timedomainviewintohowthespectrumbuildsup.EachpeakonanETCisapieceofreflectedsound
thatreachesthemicrophone.TheETCissimplylookingatamplitudeovertimebutitgivesnoinsight
intothespectrumofthatenergy.
Thekeyrequirementwithreflectedsoundsistoproperlymatethespeakersintendedresponseata
specificdistanceandaxistotheresponseasmeasuredatthelisteningposition.Typicallythismeans
keepingthespectrumofdirectandreflectedsoundsasconsistentaspossible.
ApopularapproachissimplytoanalyzethelevelofreflectionsonanETCandcomparethesetothe
directsound,settingatargetforthereflectionstobe10dBormorelessthanthedirectsound.This
analysisisnotsufficientsinceETCsarespectrallyblind(i.e.theycontainnoinformationastothe
spectralcontentofthereflectedsound)andtheauditorysystemisverydiscerninginitsrequirements
forspectralbalancebetweenthedirectandreflectedsoundsinaroom.
Thekeypointisthatspectraldistortioninthereflectedsoundsinterfereswithperceptionofboth
localizationandtimbre.Mostlistenersactuallyshowpreferenceforhighlevelsofreflectionsthatare
spectrallysimilartothedirectsound.Theyaddtotimbralrichnessandgivebodyandfullnesstoimages
withinthesoundstage.Twogoodquoteswillhelpclarifywhatishappening.Benade(FromInstrument
toEarinaRoom,JournaloftheAudioEngineeringSociety,1985)states:
"Theauditorysystemcombinestheinformationcontainedasetofreduplicatedsoundsequences(authors
notei.e.thedirectsoundanditsreflections)andhearsthemasiftheywereasingleentity,provided:

Page5of30

a) thatthesesequencesarereasonablysimilarintheirspectralandtemporalpatterns
and
b) thatmostofthemarrivewithinatimeintervalof40msfollowingthearrivalofthefirst
memberoftheset.
Thesinglyperceivedcompositeentityrepresentstheaccumulatedinformationabouttheacoustical
features(tonecolor,articulation,etc)sharedbythesetofsignals.Itisheardasthoughallthelater
arrivalswerepileduponthefirstonewithoutanydelaythatis,theperceivedtimeofarrivaloftheentire
setisthephysicalinstantatwhichtheearliestmemberarrived(authorsnotethisisknownasthe
precedenceeffect).
Theloudnessoftheperceivedsoundisaugmentedabovethatofthefirstarrivalbytheaccumulated
contributionsfromthelaterarrivals.
Theapparentpositionofthesourceofthecompositesoundcoincideswiththepositionofthesourceofthe
firstarrivingmemberoftheset,regardlessofthephysicaldirectionfromwhichthelaterarrivalsmaybe
coming"

AndWrightson(PsychoacousticConsiderationsintheDesignofStudioControlRooms,1986):
Normallyasoundperceivedbytheearsisspectrallyshapedbythepinna,torso,distance,etc.Thisaidsto
thelocalizationofthesoundsignal.Theintroductionofasimilarsignalwithdifferentcharacteristicswill
altertheperceivedlocalization

DeviationsfromthetargetsabovewarrantexaminingoneoctavebandpassfilteredETCsat500Hz,1kHz,
2kHzand4KHz.WhenanalyzingtheseETCsoneshouldbelookingforthefollowingdifferences:

BetweentheL&Rloudspeakersinthesameband(i.e.betweentheL&Rat500Hz)
Betweenadjacentbandsforthesameloudspeaker(i.e.betweenthe500Hzand1kHzfortheL
speaker)

Page6of30

FigureB.1:ThisistheLeft/RightspeakersofExampleRoom1(seepage26)whichmeetsallstandards:similarity,
smoothnessanddecay.

FigureB.2ThisroomexhibitsaraggedappearancetotheETCwhichwarrantsfurtheroneoctavebandstudies.Whilstthis
roomdoesmeetthecriteriafordecayof10dBover40msitdoesnotmeettheT60requirementsmidbandT60sareover
0.7s.Thisisactuallyseeninthegeneralreflectionpatternwhichhasnodecay(viewedbynotfocusingsolelyonpeaks).

Page7of30

FigureB.3:ThisETCmeasurement,whichisfromaleftloudspeaker,showsaraggedprofileworthyoffurtherinvestigation.

FigureB.4:FurtherinvestigationusingoneoctavefilteredETCsat500Hz,1000Hz,2000Hzand4000Hzshowsdistinct
differencesinthespectralcontent.ThesedifferencesaremoreclearlyseenwithsomeETCsmoothingseeFigureB.5

Page8of30

FigureB.5:SmoothedversionsoftheETCinFigureB.4usinga1msmovingaveragefilterandareducedverticalscale.Whilst
500Hzand1000Hztrackeachotherverywell,cleardifferencescanbeseenat2000Hzand4000Hz.Inthisroomthese
differencescanbetracedtothespeakerswhichhavepooroffaxisperformanceandthepresenceofthick,fibrouswallpaper
onallwalls.

Page9of30

C:LowFrequencyDecayTimes
Standards:

ReferenceFigureC.1.Inparticular:
o

Resonancesfrom35Hz300Hzshouldnotextendbeyond350msbeforedecaying
intothenoisefloororreachingalevelof40dB.

Below35Hzthisstandardisrelaxedto450ms.

RoomdimensionsshouldbereferencedtothestandardsinSectionGofthisdocument.
Ifthedimensionsfail,theexpectationofresonancecontrolshouldbelessened.

FigureC.1:showstheideal,minimumandacceptablerangesforLFresonance

Resonancesduetoroommodesareclearlyandmosteasilyseenonachartsuchasacumulative
spectraldecay,spectrogramorwaterfallthatshowtimeononeaxis,frequencyonanotherandplots
soundpressurelevelsasthedata.
Itisimportanttorealizethatwehearthefrequencyresponseandthatbelow250Hztherooms
behaviorismostlyminimumphasemeaningthatinpracticepeaksinthefrequencyresponse
correspondverycloselytoringinginthetimedomain.Howeverlookingataspectrogramclearlyand
Page10of30

quicklyallowsdiscriminationofthemodalresonancesandisthereforeausefulcomplimenttothe1/24th
octavefrequencyresponsemeasurements.
Resonancesshoulddecayintothenoisefloororreach40dBbefore350msotherwiseaudibleissues
mayoccursuchas'onenotebass'and'boominess'.Suchexcessresonanceisactuallyadistortionofthe
recordedwaveformaffectingtheattack,decayandtimbreofbassrangecontent.Intheworstcasesa
particularlyslowdecayingresonancecanmakeasystemalmostunlistenableasmusicalnotesatthe
resonantfrequencyseemtocomealmostomnidirectionallyfromtheroomratherthanfromthe
speakers.Inmildcasesonewillnoticethatsomenotesoninstrumentssuchanacousticdoublebasswill
soundlouderthanothersandwilldecayatadifferentratecomparedtothosearoundthem.This
unevendecaydamagesdynamicarticulation,particularlyonfastpacedmusic,andcanleadtoa
subjectivereductionindynamicimpact.
Below35Hzthetargetis450ms;itismuchmoredifficulttocontrolresonancesatverylowfrequencies
duetotheirverylongwavelengths.Furthermoreatfrequenciesbelow35Hzwearebelowthe
fundamentalfrequenciesofnearlyallmusicalinstruments.Forthesetworeasonsthedecaystandardis
relaxed.
FigureC.2:ExampleRoom1Spectrogram.ComparetoFigureC.1.IRwindowis300mswithafrequencyresolutionof3.3Hz.

Page11of30

FigureC.3:ExampleRoom2Spectrogram(seepage27).IRwindowis300mswithafrequencyresolutionof3.3Hz.

Page12of30

D:MidrangeDecayTimes
Standards:

Timetakenforsoundtodecay60dB(T60)shouldbebetween0.2sand0.5sfrom250Hz
to4kHz
T20andT30shouldbe+/25%acrossthesamefrequencybandusingonethirdoctave
smoothedbands.

SinglefigureT60measurementscannotdomuchmoreinasmallroomthantellyouwhetheraroomis
overlyliveoroverlydead.Moreusefulistolookathowsounddecaysacrossthecriticalmidrange
frequencybandsfrom250Hzto4kHzandtoexaminewhetherthespeedofthisdecayisconsistentover
time.Therearefourdifferentproblemsthatcanexistwithdecaytimestheycanbetoolong,too
short,unevenacrossthefrequencyspectrumorvaryexcessivelyovertime.
Overlylongdecaytimescanobscurelowleveldetailandthetimbreofavoiceorinstrumentinmuchthe
samewayasaroomwithahighnoisefloor.Aroomwithalongdecaytimesalsotendstosoundharsh
andbrittleandcanbeanunpleasantplacetolistenresultinginrapidfatigue.Overlylongdecaytimes
areconsideredthoseover0.5s.
Decaytimesthataretooshortmaynotallowaspacious,envelopingsoundstagetodevelopand
thereforecanbequiteboringplacestolisten.Overlyshortdecaytimesareconsideredthoseunder0.2s.
Musicdoesnotcometolifebutrathersoundsdryandsterile.Ofnoteisthefactthatacoustical
diffusorsactuallyyieldabsorptioneffectsintheprocessofdispersingenergy.Thisisimportantasa
spacewithampleuseofanappropriatetypeofdiffusormayhaveadecaytimeonthelowerrangebut
notfeeldry/sterile(likelyjusttheopposite!).Thisisaprimeexampleofthenuancednatureof
acousticalcontrolwheretheroadyoutaketoaspecificmeasurementresultiseverybitasimportant
astheendmeasurement.
Aroomthatexhibitsunevendecaycharacteristics,wherethesounddecaysmuchfasteratsome
frequenciesthanotherscanatworstsoundnoticeablyunbalancedwithadulltrebleorbloatedbass.
Unevendecayismostoftencausedbyfurnishingswithintheroomsuchasthindrapesandcarpetsthat
absorbsignificantlymoreenergyattreblefrequencies(abovearound1kHz)thantheydoatmidrange
frequencies.ProvidingtheenergyfollowsthecriteriainSectionBofthisdocument,itisactually
preferredtoretainuppermidrangeandhighfrequencyenergy.OurtargethereisforT20andT30tobe
within+/25%acrossthefrequencyspectrum.SignificantchangesindecaytimefromT20throughT60
arealsoagoodindicatorofspectrallyunbalancedreflectedsounds.

Page13of30

FigureD.1:ExampleRoom1isonthelowerendoftherangewhichisappropriateforthesecondaryfunctionofmulti
channelaudio(asopposedtohometheater).

FigureD.2:ExampleRoom2istowardtheupperendoftheacceptedrangeandwithindeviationtargets

Page14of30

FigureD.3:T20showingacceptablerangeofdeviation(+/25%)forExampleRoom3(seepage28).

FigureD.4:T30showingacceptablerangeofdeviation(+/25%)forExampleRoom3.

Afurthercomplexityisthatroomsizeandmusicalpreferencealsoplayintodeterminationofwhatthe
decaytimeshouldbe.

Onthesubjectofmusicalpreferenceswhilstmostofuslistentoawidevarietyofmusicsome
individualsnearlyexclusivelylistentoonetypeofmusic.Inthiscircumstanceitmaybeworth
consideringoptimizingthereproductionenvironmentsothatitbettersuitsthatparticular
Page15of30

musicalstyle.Ofparticularinterestisthetimeittakesforsoundtodecay.Classicalmusic
aficionadosmaybebestservedwithroomthatismorelive,withahigherT60of0.4to0.5s.
Thosewholistentonearlyexclusivelytoclosemikedmultitrackrecordingsshouldaimfora
lowertargetwithaT60approaching0.2to0.3s.

Onthetopicofroomvolumegenerallyspeakingthelargerthecubicvolumeofthespace,the
longerthedecaytimemaybe.

Page16of30

E:ConsistencyofMidrangeFrequencyResponsebetweenLeftandRight
speakers
Standard:

Inroomonethirdoctavesmoothedfrequencyresponsemeasurementatlistening
positionforeachspeakershouldbewithin+/3dBfrom250Hzto4kHz.
Eachspeakershouldexhibitadeviationofnogreaterthan3dBfromtheother

Nearlyallthefundamentalnotesproducedbymusicalinstruments,includingthehumanvoice,occur
withinarangeboundedatthetopendby4kHz.Atthelowend,fundamentalsoccurdownto35Hzor
evenslightlybelow.Onecouldarguethatahighendroomshouldthereforebeflatbetween35Hzand
4kHztomaintaintimbralaccuracyandspectrallybalancedsoundreproductionacrossthecritical
midrange.Inpractice,however,thisisverydifficulttoachieveduetothedominanceofmodal
resonancesandboundaryeffectsbelow250Hzwhichcausepeaksanddipsinthefrequencyresponse.A
sensibletargetmightthereforebestatedas+/3dBwithintherange250Hzto4kHzwhenthefrequency
responseisexaminedwithaonethirdoctavesmoothing.Inahighendroomonemighttryandgetthe
frequencyatwhichtheresponsebreaksthe+/3dBbarrieraslowaspossiblesuchas200Hzoreven
150Hz.
1/3rdoctavesmoothingisaverytraditionalwaytosmoothafrequencyresponsechartinorderto
createsomethingthatisclosertowhatwehear.Theearsresolutionisactuallyhigherthan1/3rdoctave
thoughrecentstudiesshowthattheearloudnesssummingresemblessomethingclosertoa1/4thor
1/6thfilterandeventheninbandfrequencyanomaliesaffectingtimbralresolutioncanbediscriminated.
1/3rdoctaveisneverthelessusefulforidentifyinggeneralspectraltrends.
FrequencyresponsemeasurementsshouldbeconductedforLeftandRightspeakersindividuallyandthe
responsescompared.
Sometimesdifferencesinplacementrelativetoroomboundaries,alackofleft/rightsymmetryinthe
room,poorpairmatchingoracousticallydissimilarsurroundingscanleadtoonespeakerhavinga
differentresponsetoitspartner.Thismustbeavoidedsincethisoftencausesthesoundstagetobe
pulledtowardsonespeakerconsistentlyoratcertainfrequencybands.Thefirsteffectcanbehandily
counteredbyusingabalancecontrol,assumingthatonesequipmentstillhasthisfunctionality.The
lattereffectmustbeavoidedatallcostssinceitcausesimagestowaverdependingon,forexample,
whichnoteisbeingplayedorsung.Thetargethereisforeachspeakertoexhibitadeviationofno
greaterthan3dBfromtheother.3dBisaperceivedloudnessdifferenceofaround20%.
Attentionpaidtospeakerquality,listeningroomsymmetryandappropriatetreatmentofreflection
pointsistypicallyenoughtomeettargetswithoutanyfurtheralterationstotheroomdesign.Notethat
verygoodresultscanbeachievedinslightlyasymmetricroomsaslongastheaboveconditionis
respecteditisnotnecessaryforthespeakerstobeplacedexactlythesamedistancefromtheleftand
rightsidewalls.Infactthisapproachcanoftenleadtoproblemssincethepathlengthdifferencesfrom

Page17of30

eachspeakertonearbyboundariesarethesame,causingspeakerboundaryinterferencerelatednullsto
coincide.
FigureE.1Exampleoffrequencyresponseforleftandrightspeakerthatisoutsidetolerance;inthiscasemoderateimaging
waveringwasobservedaswellasanoverallshiftinthecentralimagetowardstheleftspeaker

FigureE.2Exampleoffrequencyresponseforleftandrightspeakerthatiswithintolerance.

Page18of30

F:LowRangeFrequencyResponse
Standard:Inroomlowfrequency(LF)responsemeasurementatlisteningpositionshouldbe:

Within+/10dBat1/24thoctaveresolutionfrom20Hzto250Hzforbothspeakers
measuredtogether.
Within+/5dBat1/3rdoctaveresolutionfrom20Hzto250Hzforbothspeakers
measuredtogether.

Twomeasuresarerequiredtoproperlyvalidatethequalityofaroomslowfrequencyresponse.1/3rd
octavestudiesprovideaviewintooverallspectraltrendsequatingcloselytohowthehumanear
perceivesloudnesswhilst1/24thoctaveallowsdiscriminationofindividualresonancesthatcausetimbral
distortion.
Thelowfrequency(LF)responseofastereopairofspeakersinaroomisfirstdictatedbythespeaker
andlistenerpositionsandthenbytheacousticalcontrolthatexistswithin.Boundaryinterference
patternsarecreatedwithin~65mswhentheLFenergyofthespeakersmakescontactwiththeroom
surfacesandrecombineswiththedirectenergy.After65mstheroommodeswillimpartdistortionsif
notproperlydamped(referenceSectionD).
ToobtainthebestpossibleLFresponse:

Thespeakerandlistenerinitiallocationsshouldbecarefullyselectedbasedonproperstudyof
roomdimensionsandpracticality.Theleftandrightspeakersshouldcreateaspreadof45to60
degreesfromthecenterpointofthelisteningposition.
Afterasubjectivereviewoftheinitialresponse,acousticalmeasurementsshouldbetakento
beginthefinetuningofbothspeakerandlistenerlocations.Thesechangeswouldtypically
involvemovementsgreaterthan6.
Theremainingboundaryinterferenceissuescanbetoughertoaddress.Varyingthefixed
distancesfromspeakertoboundaryandlistenertoboundarywillreducestrong
cancellations.ItisabalancingactasonelocationthatmayofferasmootherLFresponsemay
notprovidetheoptimalmidrangeandtrebleresponse.Note:asspeakerdistancesfromthe
frontwallmayoftenbe46,aspeakerboundaryinterferencerelatednullislikelytoappear
between70Hzto45Hzrespectively.Therewouldalsobesimilartypenullbasedonthe
relationshipofthelisteningpositiontotherearwall.
Thefrequencyresponseshouldalwaysbecorrelatedtoaresonancestudyasbotharecrucialto
balanceandnaturalperceptionofLFtransients.
Moreevolvedlevelsofanalysiswouldfactorsinglespeakerdataaswellasthephaseaspectsof
theinroomresponse(excessgroupdelay).
Atthispoint,itisappropriatetolookcloselyatuseofparametricEQordigitalroomcorrection
optionstoflattentheLFresponse.OnthistopicfromDr.FloydToole,Itrytobecarefultosay
atlowfrequencies(subwooferregionbelow80Hz)smallroomsbehaveasminimumphase
systems(asitappliestoequalizingresonances).WhenpushedIgoontosaythatthisismostly
Page19of30

trueforresonantpeaksthatstandabovetheaveragespectrumlevel,butasthatlevelrecedes
belowtheaveragespectrumlevel,weclearlyhavea"signaltonoise"(minimumphasetonon
minimumphase)problem,becausethereareobviouslynonminimumphaseactivitiesinthe
regioninterferencedipsbeinganexample.Thisisrelevantbecausethebestevidenceofan
audibleresonanceisahighlyvisiblepeak,which,ifattenuatedtotheaveragespectrumlevel(i.e.
flat)resultsinasevereattenuationofringingwhichisthedesiredsituation.
Inthelargersense,everyonedesiresaflatLFresponseandnomodalringing.Simply,thisisatough
achievement.Theabsurdlylargecollectionofinterrelatedvariablesbetweentwofullrangespeakers
andtheroom(speakerdesign,speaker/listenerlocation,roomsize/constructionandacousticalcontrol
within)makesthisso.Itisuptotheindividualtodeterminewhattheirlimitsareasregardsplacement
andacoustictreatment.
FigureF.1:ExampleonetwentyfourthoctaveLFresponse.

FigureF.2:ExamplethirdoctavesmoothedLFresponse.

Page20of30

FigureF.3:ExampleRoom4(seepage29).ThenexttwoimagescompareanUNSMOOTHEDand3rdOctaveSMOOTHEDview
ofthesamedata.TheUNSMOOTHEDactuallyfallswithinstandardsat1/24thresolution.Theinterestofshowingthe
UNSMOOTHEDgraphexistsinstudyingthemanynarrowvalleyswhilethepeaksfollowaflattrend(withaslightuptilt
below50Hzthatcanbepleasing).Thesevalleysareduetoheavilyslopedceilinganglesandotherasymmetricalfactorsof
theroom.

FigureF.4showsthe3rdOctaveSMOOTHEDLFResponseofSampleRoom4.Itis+/3.5dB(wellwithinstandards).The
pointofinterestishowtogainvaluefromvariouslevelsofresolution.ThenarrownullsoftheUNSMOOTHEDresponseare
heardasflatinthissituation.

Page21of30

G:RoomSizeandConstruction
Squarefootageshouldbeintherangeof224to475sq.ft.
Roomvolume1,750to4,750sq.ft.
Theroomdimensionsshouldmathematicallyhavenocommondivisors
andnopairofdimensionsshouldsharemorethanonecommondivisor
Theuseof(2x)layersof5/8gypsumboardisadvisedforwallsand
ceiling
Irregularshaperooms,roomswithvariedconstructionmethodson
differentsurfacesandroomswithadjoiningsecondaryspacesshouldall
becarefullyconsideredforthetaskofhighperformancelistening.
BoththeSquarefootageandRoomVolumeareexcellentqualifiersofaperformancedrivendedicated
listeningenvironment.Thedesiredrangeofsquarefootageshouldbebetween225and475sqft.The
roomvolumeshouldfollowsuitwithadesiredcubicvolumeof2,250to4,750cuft.Sothen,what
shouldtheexactroomdimensionsbe?...well,itdepends(sorry,noeasyanswersexist).However,
havingatleastonedimensionbetween20and25allowstheroomslowestmodetobelowerthan
30Hz.Theceilingheightshouldbe9.5to11asthisplacesthelistenersearscomfortablybelowthe
floortoceilingmodalnull.Regardinginternalroomvolume,thetargetedrangeyieldsgreatermodal
densityextendingdowntoward50Hz.Followingtheseguideswilloffertangiblysmootherandmore
evenbassresponseandamorepureconnectiontothespeakerstrueoverallsound.
Theratioofroomdimensions(width,heightandlength)isafoundationalaspectofyourlistening
experience.Inthisregard,yourroomshouldbethoughtofasaninstrument.Properrelationships
betweentheroomdimensionswillyieldthesmoothest,mostconsistentbassresponse,period.Sorryto
reportthough,therearenofactorysetidealroomratios.Therearegoodstartingpointssuchasthe
LoudenRatioof1,1:4,1:9.Ifgivenablankpage,theroomdimensionsshouldmathematicallyhaveno
commondivisorsandnopairofdimensionsshouldsharemorethanonecommondivisor.Thegoalisto
yieldanevendensityanddistributionofroommodes.
Howthewalls,floorandceilingareconstructedisasignificantpartofthesonicsignatureofalistening
room.Massiveconcretesurfacesretainnearlyallbassenergywhilelightweightresidentialpartitions
allownearlysimilaramountstopassthroughtoadjoiningspaces.Optimalsoundqualityinthisrealmis
alotlikeGoldilocksnottoohotortoocoldbutjustright.Sonically(notfactoringisolationneeds),just
rightisadoublecourse(2x)of5/8gypsumboard.Thisyieldsenoughmassforthelowfrequency
energytobepunchyandfocusedwhendesiredbutnotsomuchthattheresonancesareoverwhelming.
Furtherbenefitscanbehadfromuseofcertainisolationhangersanddampenersastheydiminish
afterringelementsthatbuildingmaterialscanexhibitwhenfullyexcitedbysoundvibrations.

Page22of30

Conclusion
Thepurposeofthispaperistoofferclearstandardsinareasthataffecttheaudiophiletwochannel
listeningexperience.Itisimportanttorealizethatnosinglemeasurementcantellyouwhataroomwill
soundlike.Whenconsideredincombination,however,thesestandardsprovidepowerfulinsightsinto
yourroomsperformance.Itisourbeliefthatasaresultofthisarticleyouwillbemuchbetterplacedto
determineandprioritizetheacousticalareasthatneedimprovement.Further,itisclearlyshownthat
theprocessoflocatingtwospeakersandasinglelisteningpositioninaroomiscomplexandnospeaker
inandofitselfshouldbeexpectedtocompensateforallpossiblevariablesinagivenspace.
Itisnotrealistic,norisitcrucial,thatyourroommeetsallofthesetargets.Butidentifyingwhichareas
ofreproductionareimportanttoyouandoptimizingyourroomtomeetthesegoalsisanimportantstep
towardsimprovingyourlisteningpleasure.Whetheryouchoosetocompletethisjourneyyourself,or
seektheexpertcounseloftheauthorsNyalMellorofAcousticFrontiersorJeffHedbackofHdAcoustics,
youcanbesurethatthetimeandmoneyspentonoptimizingtheacousticsofyourlisteningspaceisone
ofthebestbangforthebuckinvestmentsyoucanmakeinthishobby.

Page23of30

Acousticaldesignconsiderationsfordifferentspeakertypes
Conventional(typicalforwardradiatingboxtypespeakers)
Thistypeofspeakerischaracterizedbyanonaxisandoffaxisresponsethatvariessignificantlyasthe
angleoffaxisincreases.Thisvarianceinoffaxisenergyrequiresmoreaggressiveandmuscular
acousticalcontrolatlateralandverticalfirstreflectionpoints.Absorptionofthelowermidand
midrangefrequencies(100Hz2KHz)istypicallyamusttoexperienceadynamicandcleanresponse.In
thiscase:tighterdecaytimes,lowerlevelsofreflectedenergyandmorecarefulspeakerlocationin
termsofmodalpeaks/nullsarekeystepstowardsonicexcellence.Frequenciesbelow~300Hz
propagateomnidirectionallyinthesespeakers.
ConstantDirectivity
CDtypespeakersorConstantDirectivity(alsocalledtermedcontrolleddispersion)arecharacterizedby
anoffaxisresponsethatvariesminimallyfromtheonaxisresponse.Itisnotuncommonfortheoffaxis
responseuptoaswideas40degreesinthehorizontalplanetohaveessentiallythesamefrequency
response.Itisalsotypicalthattheverticaldispersionismoretightlycontrolled.Inspeakerdesign
terms,thisisachievedbytheuseofahornorwaveguideelementinconjunctionwiththehigh
frequencyand/ormidrangedriver.Likeconventionalspeakers,frequenciesbelow~300Hzpropagate
omnidirectionally.
Dipoles
Dipolesareaquitepopulartypeofspeakerbothintraditionalelectrostatic/ribbon/planarmagnetic
configurationsandnewschoolconeopenbaffles.Dipoleshaveacoupleofcharacteristicsthatare
worthbearinginmindwithrespecttotheacousticaltargetsdefinedaboveandtheapplicationofproper
acousticdesigntorealizehighendreproduction.
1. Dipolesradiateequallyforwardsandbackwards.
2. Dipoleshavecancellationnodesperpendiculartothebaffleaxis.Duetothistheydonotexcite
roommodesalongthisaxisandhavereducedinteractionwiththesidewalls.
3. DipoleshavealowerreverberantfieldforagivenSPLatthelisteningposition(3.8dBless)
relativetoamonopoleduetotheirincreaseddirectivity.Thedistancetowhichthedirectsound
dominatesspectralbalanceisfurtherasaconsequenceofthis;thisalsomeansthatsinglefigure
T60scanbehigherfordipolesrelativetoconventionalspeakers.
4. Dipolesarevelocityratherthanpressuresources.Thismeanstheycoupletoroommodes
differently.Apressurepeakisavelocitylowandviceversa.
5. Dipolescannotpressurizearoombelowthefrequencyofthelowestmodalresonance.Thisisa
consequenceoftheirvelocitysourcenature.Insmallroomsthelowestmodalpeakcanbeinthe
30Hzrangewhichmakesasubwooferaconsideration.InadditionDipoleshavea6dBper
octavecancellationbelowthesocalleddipolepeak.Thismeansthatverylargebafflesizesand
displacementsarerequiredtoreproducelowfrequenciesathighSPLs.

Page24of30

Explainingthedivisionofthefrequencydomains
Thebehaviorofasmallroominthefrequencydomainistypicallybrokendownintoatleasttwoparts
abovethesocalledtransitionfrequencyandbelowit.Belowthetransitionfrequencytheresponseis
dominatedbytheeffectofroommodes.Above250Hztheresponseisdominatedbythespectrumof
theearlyreflections.Thisisoneofthekeyreasonswhyitisimportanttoensurethespectralbalanceof
reflectedsoundsaresimilartothedirectsound.Moresophisticatedanalysescanincludeatransition
zone,saybetween150Hzand350Hz,wheretheresponseisneithermodallydominatednorearly
reflectiondominatedbutisacombinationofthetwo.Sometimestheregionbelowaround80Hzisalso
splitoutintoanareacalledthesparselypopulatedmodalregionasthisiswhereroommodesare
spreadfurtherapart.Theactualfrequenciesdependonbothroomsizeandtheamountofabsorption
withintheroom.

Frequencies
below250HZ

Frequencies
above250HZ

Page25of30

ExampleRooms
Thissectioncontainsphotographsandshortdescriptionsofsomeoftheroomsfromwhichsomeofthe
acousticalmeasurementsinthispaperwerecaptured.
Room1
ThisisthededicatedaudiophilesurroundroomofTedBrady(moderatorofHiRezCircleatAudio
Circle).HdAcousticscompletedanacousticalrenovationofthisshowcaseroomin2010.TheSP
Tech/AetherAudioGrandRevelationsarematchedwithHdAcousticscustomdevices,RealTrapspanels
(20+)andGIKAcousticspanels.

Page26of30

Room2
KevinJoneslisteningspaceexemplifiestheresidentialsituationwithakitchenoffthetotheleftand
otheropenplanconnections.HdAcousticsperformedremoteacousticalanalysistocomplimenthis
existingGIKAcousticsmaterialswiththosefromReadyAcoustics.HisPurityAudioDesign/WilsonSasha
combinationisdescribedbyallintermsofgreatdynamicsandbeingveryrevealing.Thespeakerswere
setbyCraigHampel.

Page27of30

Room3
ThisroomatahighenddealerinSanFranciscowaspreviouslytreatedbyhangingthindrapesacross
50%ofthewallscreatinganunevenT20andT30asfrequenciesabove1kHzwereselectivelyabsorbed.
AcousticFrontiersrecommendedreplacingthedrapeswithstrategicallypositionedcombination
absorber/diffusorpanelsallowingT20andT30targetstobemetandthespectralbalanceofdirectand
reflectedsoundsmaintained.

Page28of30

Room4
JoeDaltonsroomoverthegaragefeaturestreatmentsdesignedbyHdAcoustics:customsidewall
sawtoothabsorbers,frontwallpolycylindricaldiffusorsandotherelementsworktoyieldbalanceinthis
architecturallytoughspace.ThesoundoftheWilsonSashasistrulyremarkableinthisspace.The
speakerswereexpertlysetbyCraigHampel.

Page29of30

AuthorBiographies
NyalMellor
NyalisthefounderofAcousticFrontiersLLC,acompanyspecializinginacoustic
design,acoustictreatment,roomcorrectionandsystemcalibrationforhighend
audioandhometheater.AcousticFrontiersisbasedinKentfield,MarinCounty
inNorthernCalifornia.NyalwasborninEnglandandholdsaBachelorsdegree
fromOxfordUniversity.Inhissparetimeheenjoyssnowboardingandmountain
biking.

JeffHedback
IsthefounderofHdAcoustics(formerlyHedbackDesignedAcoustics)afull
serviceacousticaldesignandconsultancyfirmspecializinginuniqueneedsof
smallroomacoustics.JeffholdsaBachelorofMusicDegreefromtheesteemed
BerkleeCollegeofMusicandbeganhisprofessionalcareerasaprobassplayer
withrecordingcreditsonmajorrecordinglabels,national/regionaltours,and
countlesssmellybargigs.HdAcousticsclientsincludeOzzyOsbourne,
LifehouseandTrevorHorn.

Acknowledgements
NyalMellorandJeffHedbackwouldliketosincerelythankDr.FloydToole,DukeLeJeune,DanFitzgerald
andBillWeirfortheirassistanceduringthecraftingofthispaper.

ContactDetails
ContactName

JeffHedback

NyalMellor

CompanyName

HedbackDesignedAcoustics

AcousticFrontiersLLC

Address

328S.AlfonteSt

1HillsideAve

Ingalls,IN46048

Kentfield,CA94904

Phone

(317)8630753

(415)2544204

Fax

(317)8630753

email

[email protected]

[email protected]

Web

www.HdAcoustics.net

www.acousticfrontiers.com

Page30of30