epa noise paper

7/29/2019 Epa Noise Paper

1/157

/J- Jd-o___-_---xJ-88Uniled Slates Office of EPA 5ro0/9-79"103Environm_nlal Protraction Noi_ AIlal_menl _nd Control November 1979Aguncy W_$hln01on , DC _04fi0Noise

_EPA Annoyance, Loudness,and Measurement ofRepetitive TypeImpulsive NoiseSources

i


2/157

_xy. ,n-EPA 55019-79-I03

ArlNOYANCE,OUDNESS,ArID _IEASUREMENTFR EP ET IT IV E TYP EIM PU LS IV E N OIS E S OU RC ES

NOVEMBER1 9 7 9

P repared by :L , C . Su the r la ndR . E . B u rkeWY L E R E SE A RCH

E l S egundo, C a lifo rn ia 90245

Prepa red for:U .S . E IIV IR ON ME NT AL P RO TE CT IO N A GE NC YOffice of Noise Abatement and ControlW ash ing ton , D .C . 20460

This report has been approved for general availability. The con-tents of this report reflect the views of the contractor, who isresponsible for the facts and the accuracy of the data presentedherein, and do not necessarily reflect the official vimls orpolicyof the EPA. This reportdoes not constitutea standard,s pe ci fi ca ti on , o r r eg ul at io n.

H


3/157

PREFACE

The United States E nvironment al P ro tection A genc y (E PA ) wascharged by C ongress in the Noise C ontrol A ct of 1972, as amended bythe Quiet C ommunities A ct of IgTB, to conduct or finance researchto inve stig ate " ... the psy chol ogic al and phys iolo gica l eff ect s of no iseon humans and the effects of noise on domestic animals, wildlife, andproperty,and the determinationof dose/responserelationshipssuitablef or u se i n d e' ci sl onm ak in g. .. " ( Se ct io n 1 4( b) (1 )) .

P ursuant to and as part of this mandate, E PA has undertaken investi-g atio ns t o det ermi ne a nd qua nti fy subj ecti ve rea ctio ns o f i ndi vidu alsand communities to different noise environments and sources of noise. Aspecific series of studies has been initiated to determine the bes:methods for evaluating subjective magnitude and aversiveness to noise onthe basis of spectral and temporal properties, and to ascertain the impor-tance of and means for including nonacoustical factors in the evaluationof general aversion to noise, The overall purpose of this line of researchis to derive a more solid basis for assessing the aversiveness of noise andthe benefits of noise control.

The aim of the investigation described in this report was to performa detailed analysis of data pertaining to potential annoyance responsesthat may be attributed to repetitive type impulsive noise. Specifically, aprogram was undertaken (1) to review and evaluate the literature onhu ma n s ub je ct iv e re sp on se to re pe ti ti ve i mp ul si ve no is e, an d ( 2) t o as se ss t he

i


4/157

need for and relative order of magnitude of a subjective impulseadjustment factor that would better define effective level in terms ofa nn oy an ce r ea ct io ns .

The report provides much useful information on the annoyance andloudness of repetitive impulsive noise. Moreover, it is expected thatthe results of the investigation will form the basis of future experi-mental psychoacoustic work to derive, if appropriate, more precise correc-tions factors or noise prediction methods to effectively account for theinherent annoyance associated with impulsive noise. E PA believes thatfurther research and evaluation of data on the subjective effects of noisewill foster the development of techniques to demonstrate additionalbenefits of noise control beyond that exhibited by currently used pro-cedures, F ulfillment of this objective awaits further study within thisseries, The results published in this report, however, do provide animportant step toward a more complete understanding of the phenomena ofhuman subjective response to noise.

The conclusions reached in this report regarding moderate level impulsivenoise are the authors' and do not necessarily reflect the opinions of theindividuals listed above. Moreover, the U. S. E nvironmental P rotection A gencydoes not endorse the findings of this investigation for use as a "correctionfactor" applicable to impulsive type noise, nor have similar correctionfactors been used by the A gency in past or current noise impact analyses.

OF FIC E OF THE SC IE NTIF IC A SSISTA NTTO THE DE PUTY A SSISTA NT A DMINISTR ATOROF F IC E OF NOISE A BA TE ME NT A ND C ONTR OLU. S, E NVIR ONME NTA L P ROTE CTION A GE NC Y

i i


5/157

ABS TP_CT

This study was undertaken to evaluate subiecfive and objective aspects of maderatelevels oFno_se from impulsive sources. The study excluded evaluation af hearing damagerlsk or annoyance from building vibration by high level impulsive noise, which werecovered by recent recommendations of the National Research Council, Committee onHearing Bioacoust_cs and B_omechanics, Working Group 69. While the study includedarlginal investigations _nta some of the objective aspects of impulsive noise, o detaTledreview of the llteroture on the subjective aspects was emphasized. Based an thls availableHterature, the annayance and loudness from a wlde variety of repetit ive impulse noiseswere evaluated These results were applied to the evaluat_an of _mpulslve noise froma number of specific noise sources. Based on the mast pertrnent literature, it _s ten-tofively cancluded that a subiectlve impulse correction factor of 7 dB applied to theA-werghted equivalent sound levels of these types of repetitive impulsive noise sourceswould better define their effective level _n terms of annoyance reactions. No addlt_onalcorrection is identlfled at this time for crest level or repetrfion rote. Research an sub-jectlve correction factors for hellcopter blade slap is also reviewed and potentialreasons for the smaller subjective correction factors (i.e., 0 to 6 dB) for annayanceresponse to this type of sound are discussed. It _s recommended that refinements to thissubiectlve correcfian factor be based an the use of standard loudness calculation methods{Stevens Mark VII or Zwlcker) mcdffled to include provision for o shorter time constantto reflect subjective response to short duration impulsive sounds.

The study also included a brief experimental evaluation of the measurement of owlde variety of simulated repetitive _mpulsive-type signals vorylng in duty cycle, repeti-tion rate, pulse frequency_ and ratio of peak _mpulse signat level to continuous backgroundnoise level. When repetitive impulses are measured using maximum values of A-weighted(slaw) readings on an Impulse Sound Level Meter, no ob[ectlve correction is necessary inorder to measure, with an accuracy of _ 1.5 dB, the equivalent sound level (Leq) of thewide variety of impulsive signals investigated.

i i i


6/157

ACKNOI#LEDG_ENTS

A program to develop techniques for evaluating the noise from selected impulsiveno_sesources was inltloted in December 1975 for the Office of Noise Abatement and Control,U.S. Envlronmental Protection Agency. Mr. Jeff Goldsteln served as technical monitor.

The problems addressed in this study encompc,ss areas in psychoacousHcs and acousti-cal measurement technology which have received intensive study by many investigators formany years. The authors wish to thank the following individuals who, during the conduct ofthis study, pfovlded much helpful information from research which they or their col leaguesh av e c on du cte d.

Dr. Robert Gales, Naval Undersea Center, San D_ego, CaliforniaDr. William J. Galloway, Bolt, Beranek and Newman, Canaga Park, CaliforniaMr. B.W. Lawtan, NASA langley Research Center_ Hampton, VirginiaDr. O. Juhl Pedersen, Technical Universlty, DenmarkDr. D.W. Roblnson, National Physical Laboratory, EnglandDr. Bertram Scharf, Northeastern Unlvers_ty, Boston, MassachusettsDr. Paul Schomer, U.S. Army Construction Engineering Research Laboratory,Champaign _ IlllnoisDr. Milton Whitcomb, Natlonal Academy of Soleneest CHABA, Washington, D.C.Dr. Robert Youngt Naval Undersea Center, San Diego, California.The valuable contr ibutions which were made to Appendix A by Dr. Mark Lee, presently

of the Jet Propulsion Laboratory, Pasadena, California, and to Appendix B by Mark Montroll,are also appreciatively acknowledged.

iv


7/157

TABLEOF CONTENTSPage

i.O INTRODUCTION .......... I-I2.0 SELECTIONOF A BASELINEMETRIC ..... 2-I

2, 1 DeFinition oFimpulsiveNoise 2-12.2 Baseline Noise Metric 2-10

3.0 SUBJECTIVERESPONSES TO iMPULSIVENOISE 3-I3.1 Loudnessr Noisinessof impulsive Sounds . 3-2

3.1.1 A Model Farthe Hearing Process 3-23.1.2 ExperimentalData. 3-43.1.3 Subjective Responseo Impulsive PureToneSounds 3-83.1,4 Subjective Responseo BurstsoFNoise 3-173.1.5 LoudnessVersusNoisinessoF ImpulsiveNoise . 3-213.1.6 Subjective ResponseaComplex impulsiveSounds 3-27

3.2 Annoyanceand Other Subjective Responseo impulsiveNoise. 3-333.2. I AnnoyanceResponseo Impulsive Noise 3-343.2,2 Helicopter BladeSlap Noise 3-393.2.3 LoudnessVersusAnnoyanceof ImpulsiveSounds 3-473,2.4 Other Subjective EFfectsel: Impulsive Noise 3-50

4.0 CONCLUSIONs SUBJECTIVECORRECTION FACTORS FOREVALUATION OF IMPULSIVENOISE 4-I4.1 Subjective CorrectionFactorAs 4-1

4.1.1 Subjective Correction FactorsBasedon LoudnessResponseData for Toneand Noise Bursts 4-I4.1.2 Subjective Correction FactorsBasedon Measured

Loudnessof Real ImpulsiveNoise SOurces 4-2

V


8/157

TABLEOF CONTENTS (Continued)P o _ e

4. I. 3 Subjective CorrectionFactorsBasedon Annoyance . 4-34.1.4 Summaryof MethodsForComputingthe SubjectiveCorrectionFactor _ 4-8S

APPENDIX A OBJECTWEMEASUREMENTOF IMPULSIVENOISE A-1APPENDIX B ISO ROUND ROBIN TESTS B-1APPENDIX C FREQUENCYSPECTRAOF REPEATEDTONE BURSTS C-1REFERENCES

Part A - Annoyanceof Impulsiveor Fluctuating Sounds R-IPart B - Noisinessand LoudnessFImpulsiveor FluctuatingSounds R-5PartC - Detection or Perceptionof ImpulsiveSounds R-IOPort D - SpeechInterference R-11Port E - Sleep Interference R-12Part F - Hearing Damago R-12Part G- MoosurernentoFImpulsiveSounds R-14

kJ, .

' Vi


9/157

LISTOF TABLES

Table PageNo. No.! Typical PhysicalParametersof FourReal Sourcesof ImpulsiveNoise 2-92 Index of ExperimentalStudiesan Loudness/Nolsinessof ImpulsiveorFluctuating Soundsndicating ExperimentalVarlables Investigated 3-63 Description of Naturafly Occurring ImpulsiveSoundsasComparisonSignals 3-29in EvaluationExperimentby Fidell and Pearsans4 A Summaryof Literatureon AnnoyanceResponseso ImpulsiveNoise(ExcludingStudiesfor HefleopterBlade Slap) 3-355 Summaryof RecentStudiesof Helicopter BladeSlap Noise IncludingSumrr_ryoESubective CorractionFactor for Impulsiveness 3-406 Comparlsonof Several PredictedSubjectlveCorrectlon FactorsforAnnoyanceApplied to the Four Impulsive Noise Sources 4-77 Sumrnawof SubjectiveCorrectionFactor (As)EsHmatedrom ExistingMethodsor Data, dB 4-B

vii


10/157

LISTOF FIGURESFi_lure Pa_Ie

1 Examplesof Time History Envelopesof Nonimpulsive and ImpulsiveSounds 2-22 Examplesof Time Historiesof the Instantaneous'PressureromImpulsive Sources 2-43 Physical Parametersof a Typical ImpulsiveSound 2-84 Conceptual Illustration ofAuditory ProcessoShowCharacteristicResponseTimes(I') in VariousElementsWhich Govern the Dynamic

Responseof the Ear to TransientSounds. 3-35 Measuredand Normalized ValuesForChange in Signal-to-Noise Ratioof Single ToneBurstasa Function of BurstDuration B-96 Rangeof MeasuredLouclnessLevel - DurationTradeoffReportedfromVariousStudiesto Indicate PossibleRangeof Uncertainty in PredictedLoudnessof 20 msPulse 3-127 Subjective Correction L_sfor R,:peated1000 Hz ToneBursts B-13B Subjective Correction ,', for RepeatedPureToneBursts_PulseDuration

20 ms, Repetition Rate--s25/Secondasa Functionof (a) ReferenceIntensityt L (Ref) at 1000Hz and (b) Frequencyfor the80 clBReferenceLevel eq 3-14

9 Time History and FourierSpectrumof a Typical ImpulsiveSignal 3-1610 SubJectiveCorrectionFactor for Loudnessr NoisinessResponseoShort Burstsof Noise BandsRelative toa ReferenceNoise 3-1811 Difference Betweenthe Leqof a Repetitive Noise BurstSuperimposedona SteadyBackgroundNoise and the Level of a ContinuousNoisewhich SoundsEqually as Loudasa Functionof theRatio of the BurstLevel to the BackgroundLevel 3-1912 Time-Amplitude SequenceDiagramof theStimulusPresentation 3"_2213 Resultsof Experiment] 3-2313a Comparisonof Loudnessnd NoisinessversusRepetition Rate for aBurstTime Fraction of 0.063 3-25

v i i i


11/157

LISTOF FIGURES(Continued)F.i_ure Page

14 PreliminaryValidation of"AssessmentMethods 3-2615 Comparisonof LoudnessCelculotlon Methodsfor Trlongutar Transients 3-2816 Comparisonof Time-lntegrated Measuresof the Impulsive Noiseswith

the SameMeasureof the Equally NoisyReference Sound 3-3017 CorrelaHonof JudgedDegreeof Helicopter Blade Slap VersusCrestLevel 3-4318 illustration fromTwo Groupsof HelicopterBlade Slap Data"l_atRankOrder ofAnnoyanceDoesNot Correlatewith Judged ]mpulslveness 3-4319 Comparisonof NoiseLevelsfor EqualAnnoyance VersusEqual Loudness 3-4820 Typical Transmissionesponsef the Outer and Middle Ear 3-5121 "l'he1968CHABADamage-RiskCriterion for ImpulsiveNoiseExposureanda ProposedModification Fora Nominal Exposureof 100impulsesPer Dayat Normal Incidence 3-5422 Comparisonof Measuredand EstimatedValuesof the SubectiveCorrectionFactor_ asa Functionof Crest Level 4-10S

iX


12/157

1.0 IN 11_ODUCTIONUnder the mandateof the Noise Con_ol Act of 1972, the Environmental

Protection Agency is charged with taking steps to abate sourcesof noise potentially,detrimental to the public health and welfare, implicit in this is the need to establishthe meansfor evaluating and monitoring the noise from impulsive noisesources.

Thisreport excludescons_deratlonof humanresponseto'and measurementofhlgh level impulsivesoundssuch assonic booms, weaponsfire, or quarryblasts. Thelatter topic hasbeen the subject of recent recommendationso the Federal Govern-ment by Working Group 69 of the National ResearchCouncil, CommitteeonHearing,fiioacousticsand Biomechanics(CHABA). With this/imitation in mind, a researchstudy wascarried out to develop an interim methodfor the evaluation of moderatelevels of impulsive noisebelow hearingdamage risk levels. The methodwasto becompatible with theexisting methodologycurrently in useby the EnvironmentalProtectionAgency (EPA) forevaluating community noise impact. The investigationwasdivided into threebasic elements:

1. Selection of a baselinemetric for evaluating impulsivenoisetowhichsubjective and objective correctionfactors* couldbe app/iedasnecessary.

2. Reviewandevaluation of the literature onsubjective effectsofimpulsivenoisewith emphasison date relating to annoyance,noisiness, or loudnessof repetitive typesof impulsivenoise.

*'l_roughoutthis report, the term "subjective correctionfactor" is usedas a convenientlabel for the difference between Ihe subjectively effective and objectively measuredvalue of Ioudness_noisinessor annoyanceas defined in the text. It is not intendedto imply that the velues cited Forthese "correction" foclorscanbe used withoutcareful considerationof their validity andapplicability Forpractical evaluationofreal Impulsivesounds.

1 - I

il


13/157

3. Basedon this review, the developmentof a suitable method to accountForsubjective (annoyance)effects of impulsive noiseutilizing suitablemeasurementmethodsand curren!ly available instrumentation.

Thisreport presentsthe resultsof this investigationin the following sequence: Section 2 discusseshe selectionof the baseline noisemetric used

throughoutthe study. Section 3, the heart of the report, reviews the literature in detail on

loudness,naislness,andannoyanceresponseso impulsivesounds. Othersubjective effectsare alsobriefly covered.

Section4 summarizeshe overall findingsin termsof the differentialsubjective responsebetween impulsiveand nonimpulslvesounds.

Threeappendicesare alsoincluded, covering: Appendix A - Objective factorsinvolved in the measurementf

impulsive noise. This includespresentationof resultsof a laboratorytestof variousnoisemetricsobtainedfroma precision impulsivesoundlevel meter whenappliedtoa wide range ofartiflcially-generatedimpulsivesounds.

e AppendixB - Summeryof theresultsof on internationalRoundRobinteston responseo and measurementf impulsivesoundsecentlycenductedby the InternationalStandardsOrganization.

AppendixC - Frequencyspectraof repeatedtime bursts. Thisappendixbriefly illustratesthe spectral contentof various idealrepetitive toneburstswhich roughlyapproximatesomeimpulsiveseunds.

1 - 2


14/157

2.0 SELECTIONOF A BASELINEMETRIC2.1 Definition of Impulsive Noise

Soundscan be defined asimpulsive when they exhibit someform of rapid andsubstantial variation _nthe envelope of the time history of the instantaneouspeakpressures. Th_senvelope can be visualized as a llne connecting the instantaneouspeaksof a noisesignal as measuredon a h_gh-speedoscillograph. Examplesofenvelopesof impulsiveand nonlmpulsivesounds,illustrallng thisqualitative definition,are shownin Figure 1. Figure la showshe envelope of peak pressuresor Fairlysteadysoundsfrom a stationary no_sesourcesuchas an electric motor runningat con-stant speed. Figure lb showsa noisewith a noticeable fluctuation of the envelope. Thismay simply be called an unsteady or fluctuating noisesuchas froma streamof highlyvariable traffic.

The firststep in defining a baselinemetric for the impulsivesoundsconsideredin thisreport was to classifyall typesof impulsive-like soundsnto categories. Asillustrated in the figure, mosttypesof impulsivesoundsit intotwo basiccategories.Figureslc and ld showenvelopesof the time history for soundsn thesetwocategoriesthatare clearly impulsive Figure lc illustratesa single impulsesuchas froma quarryblastand Figure ld showsa repetitive impulsivenoisesourcesuchas from an unmuffledrock _,r[llor drophammer._'

Thereare clearly other exampleswhich fall somewherenbetweenthe timehistorycharaoteristicsshownhere. Forexample, the enveloperepresentingthe timehistoryof an aircraft may look quite slmilar to that of the singleimpulsivesoundexcept that the tlme scale is stretchedout to manysecondsinsteadof hundredthsofa second. However, inorder to take advantageofany usefulresearchthatcould berelated to impulsivenoise, investigationson subjective reactionstoall of the lastthreeexamples illustrated in Figure1 were groupedinto threecategoriesaccording tothe typeof soundas Follows:*The lati_"r is a wheeledvehicle equippedwith a hydraulically operateddrophammerand |s usedfor demolition of roadsurfaces.

2 - I


15/157

t r_ea.--ta) Steady Sound

I I sec..._b) Unsteady or Fluctuating Sound

I_.,0ms-Ic) Sfngle ImpulsiveSound

d) Repet_tlve Impulsive SoundFigure I. Examplesof TimeHistory EnvelopesoFNonimpulsivo(seea, b)and fmpulslva(seec, d) Sounds

2 - 2


16/157

I - Repetitive ImpulsiveSoundsi! - Single ImpulsiveSounds

I!! - UnsteadySoundsThisreview of impulsivenoisesisnecessarilybroadand potentially applicob/e to

a wide rangeof moderateto low level impulsivesounds. To illustrate the conceptspre-sented in thisreportpertainingto loudnessand annoyanceof repetitive _mpulsivenoises,four particular sourceswereselectedas typical of impulsivecommunitynoise. Theseare:

Truck-MountedGarbage Compactors1 DropHammers

Two-Cycle Motorcycles RockDrillsClearly, someof thesesourcescangenerate impulsive noiselevels which may

representa hearingdamagerisk to the equipmentoperatoror an immediatelyadjacentbystander. However, hearingdamageas_cls of impulsivenoise are notconsideredin any detail in this review. Undercertain operating conditionsor with suitablenoisecontrol features, thesenoisesourcesmay notemit whatwouldbe called impulsivenoiseaccordingto ourqualitative definition (i.e., rapid and substantialvariationin the envelopeof the peak pressureime history). However, accordingtoour threecategoriesabove, all fourof thesesources,when9eneratln_ impulsivesound,willfall into Category!, i.e. s sourcesof repetitive impulsivesounds.

Typical time historiesof the instantaneoussignalsfor eachof the abovesourcesare illustrated in Figure2.* Forgarbagecompactors,ignoring the steadynoiseof thepower sourceusedfor itsoperation, the impulsTvenature of compactornoisewillconsistof randomor irregular impactsof metal againstmetalsothat the term "repetitive"must, in this case, be interpretedas including suchan aperiodic or randomrepetition.Forthe other three sources,however, one can expect that underanygiven operatingcondition, the repetition ratewill be fairly constantso that the envelopewill exhibita definite periodicity. It shouldhe pointedout that repetition ratesof concerninthis report will fall belowthe auditoryrange, that is, belowabout 20 Hz.

The time historiesshownin Figure2 were obtained from a small samplewithineaoh sourcecategory. Theyare notnecessarilyepresentativeof all equipmentt_at fall within thosecategories. 2 - 3


17/157

-_ 200 ms I"

.t_m_a) CommercialGarbage Truckwith Compaclor

Figure2. Examplesof Time Historiesof theInstantaneouspressurefromimpulsiveSources

2 , . 4


18/157

b) Drop Hammer

Figure2 (Continued)

5-5


19/157

='J2.5 ms 14c) Two-Stroke Motorcycle

F_gure2 (Continued)

2 - 6

I__l ilrll


20/157

f ,0ms"d) Rock Drill

Figure 2 (Concluded)

2 . - 7


21/157

A typical train of impulsive soundss illustrated in Figure 3. Thef lve physicalparametersimportantFordescribingimpulsivesoundare defined Forpurposesof thisreportas Follows:

Crest Level - Thedifference in soundpressureevel betweenthepeakand rmslevel c_fthe noise. Fora backgroundnoisewith a normal(Gaussian)dislrlbut_onof instantaneousressure,the peak pressuremaybe consideredas the value at about three standarddeviatlonsabove thermsvalue. Thispeak, which ideally is exceeded only 1percentof the time ForGaussiannoisetwill be about 10 dB higherthan the rmsvalue. Thus, the crest level shouldnormallyexceedabout 10dBbeforea noiseisconsideredimpulsive.

Duration- Theamountof tlme that theenvelope of the instantaneouspressureexceedsthe rmsvalue.

I Period(if repetitive) - The time durationbetweentwo successiveimpulsesin a train of impulses.

Spectrum- TheFrequencydistrlbutlonof acoustic energyin the impu]so. Rise Time - The time requiredfor the impulsetorise Fromthe back-

groundnoisetothe peQk.

-7-

Crest Level_ _ __RiseTime -- D: Duration

P: PeriodP

Figure3. PhysicalParametersofa Typical ]mpu]slveSound

2 - 8


22/157

Representativevalues for these impulsive noiseparameters for the two-stroke motor-cycle, the drop hammer, rock drilJ, and truck-mounted garbage compactor are listedin Table 1.* For thesesourcesof impulslve noise, the crest level lles between 13 and 30 dB,the duration varies from several m_lllsecondsto half o second, and the period varies from 10milliseconds to 1-1/2 seconds. A frequency rangeof 200 Hz to 2 kHz coversmostof theacoustic energy of the impufslve noise. This table provides a general indication of"themagnitude of the parameterswhich define the general physical characteristics of the impul-slve noisesourcesconsideredin this report. However, this range of parameters, in fact__ncludesmany other impulsiveno_sessothat research_ntosubiective responseo all of thesecan be applied, in part, to the evaluation ofsubiectlve responseo the fourparticularsourcesidentified _nTable I.

Table ]Typical PhysicalParameterso Four Real Sourcesof ImpulsiveNoise

Peaks_nimpulsive Crest Pulse Repetltian Frequency TypicalNoise Level Duration Period Spectrum Rise TimesSource dB ms ms k Hz ms

Two-Stroke 13 2 - 20 30 - 100 0.30 - 2 2MotorcycleDrop 30 300 1500 0.25 - 1 10HammerRockDrill 19 10 50 0.040 - 0.400 2Truck-Mauntec 19 500 5000 0.200 - 1 50GarbageCompactor

*The values listed in Table1were measuredfroma smallsamplewithin each sourcecategory. Although there is noreasonto suspectthat the values listed are atypical,the reader shouldapplycaution in generalizlng the conclusionsof th_ss'tudyasnecessarilyrepresentativeof all equipment that fall wHhlneach sourcecategory.

**Although selectedasa repetitive impulslvenoisesourcefor purposesof thlsanalysis,recent in_'ormatlonaspresentedin EPAReport No. 550/9-79-257, RegulatoryAnalysisof the NoiseEmissionRegulatlonfor Truck-MountedSolid WasteCompactors, indlc:etesthat th_sfeature may notbe necessarilycharacteristic of the majority of truck-mountedsolid wastecompaotionunits.2 -9


23/157

2.2 BaselineNoise MetricSomesortof baselinenoise metric is necessaryfar evaluating thesevarious

impulsivesounds. Thisbaseline metric shouldbe: (1) reasonablyunambiguous,2)measurablewith precisionlaboratory equipment, (3) measurablewith standardsoundlevel metersin the field with suitable correction factors, (4) compatiblewith theday-night soundlevel (kdn) or the equivalent (energy average) soundlevel(Leq) metric, and(5) able to provide a foundationfor application of subjectiveimpulsive noisecorrectionsto allow comparisonf thesubjective responseo impulsiveandnonlmpulsivesounds. Thebaselinemetricsapplicable to the Category ! impulsivesoundscould takeone of the following alternate Forms.

SoundExposureLevel - The tlme-lntegrated measureof theA-welghtedsoundevel is identified by the symbolLS.

e Equivalent Sound Level - The equivalent soundlevel is the energy-average of the integratedA-weighted soundlevel overa specifiedobservationime Tand is identified by the symbolkeq.

PeakSoundLevel - ThemaximuminstantaneousA-welghted soundpressureevel duringa given observationtime is identified by thesymbolLApk.

PeakSoundPressureLevel - Themaximumnstantaneousnwelghted(linear) soundpressurelevel duringa givenobservationtime isidentifiedby the symbolLpk.

All of thesemetricsare essentially unambiguousquantities measurablein thelaboratoryand potentlally measurableby someof the advancedintegratingsoundlevelmeters. Measurementof the peak levels (LApk or Lpk)with soundlevel metersequippedwith a peak-hold posltionis straightforward,providingthe rise-time of

2-10


24/157

the signal _sgreater than 50/_secs. Th_scorrespondso an upper Frequencylimit of20,000 Hz for significantenergy in thespectrumof the impulsive sound.

]ntenfionally excluded Fromthe candidatebaseline metricsare theotherquantities measurableon a soundevel meter. Thosewhich will be consideredlaterfor application to measurementof impulsivesoundsnclude:

SlowSoundLevel - Theexponentlal-averagedA-weighted soundlevelmeasuredwith a nominaleffective (squaredpressure)tlme constantof1second, identified, for this report, by the symbolLAS.

SoundLevelor FastSoundLevel - Theexponential-averagedA-weighted sound level measuredwith a nominal effective timeconstantof 125 ms, identified, Forthisreport, by thesymbol LAp"

e ImpulseSoundLevel - Theexponentlal-averaged soundlevel measuredwith o nominaleffective time constantof 35 ms,. identified by thesymbolLAI.

Other noisemetricscould have been considered, such as measuresof statistical distribution, L , where x is the percent exceedencelevel, or noisexpollution level (LNp)which attemptstoaccountfor subjective reaction tofluctuationof a noise. Thesewererejectedas not being directly compatiblewith current EPAnoisemetricsandare not readily measurableonstandard soundlevel meters.

Returningto the Fourcandidatebasellne metrics, the lasttwo measuresofpeak level maybe rejected at the outsetas unsuitable becausethey Fall to fit directly

; into EPA'stime integratedmeasuresot'nolse, namely, day-nightsoundlevel Ldna,3dequivalent level (Leq). In order to makea final choice, it isnece_ary to considerthe generalnature of the noisesignaturesthat may be Tnvolved. For example, thetypical no_seexposureofan individualat any one place togarbagecompactornoisemight consistof severalminutes of exposuretoa relatively randomseriesof impulsesgeneratedby the clankingtogetherof garbagematerialsas they are compacted, super-imposedover the risingand falling humof noise Fromthe enginewhichdrives the compactor.

2 - 1 1


25/157

The duration oFthe exposurecan only be roughly estimatedand will vary w_delyfromone site to anotherand Fromoneday to the next. The soundexposure level of sucha varying noiseexposurewould also vary accordlngly, making it difficult to utilizefor realistic noise evaluation or certification unlessone observation t ime were arbi-trarily fixed. In this case, however, an equally useful measurewould simplybe theequivalent (or energy average) soundlevel (Leq) during the measurementperiod.

]n contrast, duringa passbyof a motorcycle, theonly unambiguou=energy-related measureof the no_sesignaturereceived by a nearbyobserverwould be thesoundexposurelevel (Ls). It wouldbe possTh[eo normalizethe soundexpo-surelevel by a standarddurationof, say 10 secondsto provide what wouldamount to theequivalent soundevel over I0 secondsi.e., Leq(10 sec))with thesameenergyasthe actuat event. On theother hand, if noisecertification testsofmotorcycleswere tobe applied to stationaryvehicles, the equivalent soundlevel (Leq)during the observationperiod wouldbea logical baselinemetric.

For the drop hammeror rock drill, a typical nolses_gnaturecould consistof a relatively long periodof exposure, on the order of an hour or marewith manyperiodsof moreor lesscontinuousexposureto the repetitive impulsive sound. In thiscase, agaln, theequivalent soundlevel (Leq) during the observation periodappearssuitable as the baseline metric.

Thus,with the one exception of noiseexposureto single events, which areconveniently definedby the soundexposure level, it appearsthat the equivalent soundlevel (L ) is the logicalchoice fora baselinemetric for the impulsivesourcescon-eqsideredin thisstudy.

The A-weighting inherently incorporatedin thismetric isexpected to providea moreaccurate or a moreconsistentcorrelation with humanresponseto low levelimpulsivesoundshanwouldbe providedbya nonwelghted(linear) soundpressureevel.As will be discussedater, this observationis alsoconsistentwith the observedloudnessor noisinessof low level sonicboomsounds. Thesehavebeen showntocorrelate best

2-t2


26/157

with Frequency-welghtedmeasuresi.e., loudnessin phons)of the sonicboomenergyspectrumwhich deemphaslzesthe low frequenciesas doesA-welghtlng.*46' 551 60

It remainedonly to define theobservationtime uponwhichthe averagesoundlevel will be based. For the generalcase, the equivalentsoundevel (Leq)over anobservationtime T will be definedas

T

,j Leq= 10 IOgl0 ( PA

wherePA(t) = instantaneousA-weighted soundpressuret time t, Pa

Po = reference pressure20 h_Pa), andT = observationperlod_sec

Farprediction of the day-nlghtsound level (L.), the L for the impulsivedn eqsoundis evaluated far the daytime (Ld) - 0700-2200_ and for nighttime (kdn) - 2200 to0700 hours. 111enormal 10dB penalty factor would be imposedon /dn Forthe baselinemetric, but the possibility of increasing this For the potentially even greater annoyanceat night of impulsive soundscan be left asan option to be defined upon the basisofexamining the available information on sleep interference Fromimpulsive sounds.

Forapplication to defining the k of"repeatedsingle events, the some techoe qn|que employedFarspecifyingaircraft soundexposurewill be usedin the formLeq LS+ 10 logN - 10 log [ T/t'l dB (2)

*Note that for _impulslv.e. sounds,suchas fromquarryblastsor artillerytC-weighted levels appear topredict communityresponsequite well.21t 147

2 -1 J


27/157

whereLS = soundexposurelevel ofone event, dBre 20p.Pa . secN = numberof eventsduring the time TT = observationperiodin secondst = referencetimeof I second

TheobservationtimeT toapply in the measurementf theequivalentsoundlevel will dependon the application, ranging froma minimumof 1 secondcorre-spondingto the durationof referencesoundsoften usedin laboratoryevaluation ofimpulsivesounds), to 1 hourforan hourlyequlva!ent soundlevel (Leqlh)), to 15hoursfor the day soundlevel (Ld) - the energyaverageduringthe hours0700 to2200.

In summary,then, thebaselinemetric usedin this studyfor evaluationof.impulsivenoisewill be theA-weightod equivalent soundlevel (Leq) measuredovera time tobe specifiedas appropriatefor each source. Thisprovidesa baselinenoisemetric that is compatiblewith the existing methodsdevelopedby EPAfor evaluationof noise impact.* Byprovidingadjustmentfactors(nominally identified hereinascorrectionfactors) to the Lq toaccountfor any subjectiveeffects andmeasurementerrorsfor impulsive noise, it will be possibleto properlyinclude impulsivenoisesinEPA'sevaluationof environmentalimpactof impulsivenoisesources. Thismetric isalso consideredappropriateforapplication to eachof the three categoriesof soundsdefined earlier= (a) Category] - Repetitive impulsiveSounds,(b) Category|] - SingleImpulsiveSounds,and(c)Category Ill - Unsteadyor Fluctuating Nonlmpulsi_,eSounds.

*U.S. EnvironmentalProtectionAgency, "informationon Levelsof EnvironmentalNoise Requisite to ProtectPublicHealthand Welfare with an AdequateMarginof Safely." EPAReportNo. 550/9-74-004, Mo_ch1974.

2-14


28/157

'_ 3.0 SUBJECTIVERESPONSESTO IMPULSIVENOISESubjective responsesf people to noise can beconveniently groupedinto three

general (and overlapping) categories: Health-Critical Responses

- Hearing damageLong-termmedical or psychologicaleffectsother thanhearing

damage Activity or Behavioral-lnfluence Responds

- Speech interference- Sleep interference- Task interference

Attitudinal or Judgment-lnfluence Responses- Annoyance responses- Loudnessor noisiness)judgments

The primary concern for subjective responsesn this report is in the lastcategory(i.e., attitudinal or [udgmeni responses)endtherefore that category is the only catggo_that hasbeen reviewed in depth. An extensive bibliographyhasbeencompiled, however,onmostof the above categoriesand is included in the Referencesectionat the endofthisreport. For convenience, thebibliography is arrangedchronologicallywithineachof five general subjects: Part At Annoyance of Impulsivoor Fluctuating Soundl;Part B_Loudnessor Noisinessof Impulsiveor Fluctuating Sounds;PartCt Dal_ctionQrPerceptionof ImpulsiveSounds;PartD, Speech Interference; PartE, Sleep InterFerence;

andPartFt Hearing Damage. An additional subdivisionPart G, for the referencesonmeasurementof impulsivesoundoisalso included in this bibliography. While all of thesourcesare listed in the bibliogrophyt for convanience_only the principal onesof con-cern for this report are cited as references in the mainbodyof the text.

3 - 1


29/157

3.1 Loudnessr Noisinessof ImpulsiveSoundsAswill be shownlater in Section3.2t a correction to L to accountfor thee qannoyanceof impulsivesoundscan rangeFromapproximately5 to 15 dBt dependingon

the correctionmethod. Clearly_ sucha wlde range of correctlon,Factorsisof littlevalue so that a moreprecisemethodforselectinga subjective correction factor isdesired. The extensiveliterature on loudnessor noisinessof impulsivesoundswastherefore reviewed emphasizingexperimentalresultsasa more reliable basis,at thispoint, for assistingn the selectionof a subjective correction factor. In additlonathesebasic experimentalresultson responseo transient soundsare expectedto assistin defining optimumwaysto monitor impulsivenoise. Followingthe reviewin thissectionof the avai]eble experimentalresuItson loudnessend noisinessof impulsivesounds_informationrelated to the annoyanceof suchsoundsand comparisenofannoyanceand loudnessor noisinesss consideredin the next section. First_howevertit is helpful to considera simplifiedmodelfor the auditoryprocessasa frameworkforexaminingthe data relative to impulsivenoiseresponse.

3.1.1 A Model Forthe HearingProcessA s_mplifiedconceptualdiagramof theaudltory systemis illustrated in Figure

4 to assistin defining theprincipal featuressignificant in thisstudy. As indicated inthe figure, characteristicresponseimes for the "acoustic" partsof the auditorychaln(i.e._ up to thepoint in the innerear wherespectrumanalysisoccurs)are muchlessthan the "RC" tlme-constant inside the lastbox wherethe overall detection, integration_. 146and recognition of soundsignalsis assumedo occur. Evenconslderingthelowestreported value for this tlme-constent, it isstill morethan me ordersof magnitudegreater thanfor the earlier partsof theauditorychain whichmustbe able to respondto instantaneouspressurechangesat rotesup to 20t 000 timesper second(_"= 50 _sec).The "RC" tlme-constant, ontheother hand, only limits the abi/ily to track theenvelope el_a sound. Thus,experimentalstudieson responseof humansto transientroundshave focusedmoreattention on thispartof the hearingprocessand haveutlllzed the RCsmoothingilter concept illustratedasone of the ways to emplrlca[lymodel the results. We will considertheimplicationsof the model illush'ated in

3 - 2


30/157

p .........................................[n_r!enl Sound Beflaled I BrainFlora Nearb_ _oundoeies I Analysis of Signal

/ 1 Into (Ci*ial1 Dondl''4 / /

D;f flaction ( Spoclr ur'_ Prace_lnUBy Hood Analyzer) and Pattern

Smoorhlng Filter Recognition

Tnclder*t Sound T=5OZ_s T'3Su= t [ "-D;_ecrlv f_o,e , D_l_ctlon and J X

L........ -Ir=RC - 13- IOOOms

Figure 4, Concep_'ual Ir[ustrat'ion of Audltory Processto Show CharacferlsHc ResponseTimes ('r) TnVarious Elements Which Govern the DynarnTcResponseof theEar to Transient Sounds. i',Jote: the Range for the Value of. "RC" Time-Constant Reflects the Exl"reme Rangeof"@bserved Values. (After Bruel,Reference 146)


31/157

Figure4 again later, but first let usexaminetheexperimentaldata on /oudnessandnoisiness.3.1.2 ExperimentalDqta

The independentand dependentvarlablesinvolved in the noisinessof impulsivesoundsmaybe categorizedas follows:

independentVariables (The Stimulus)- SignalFormat

Repetition, Singleor Multiple [rnpuTses Signal SpectrumTone, Narrow BandNoise, Complexor Wide Band

[mpu[slveNoise (the compleximpulsencludes the type of realimpulsivesoundsof concernin thisreport)

- SignalCharacteristicsVaried Pulse Duration PulseFrequency PulseRepetitionRate (for Repeatedimpulse) Spectrumof Total Signal Riseand DecayTimeaf Pulse PhaseoESignalComponents Ratio oFPulseSignal Level toany BackgroundNoise Durationof Total Exposure Methodof SignalPresentution(includingsoundFieldcharocterlstios

for loudspeakerpresentation)

3 .- 4


32/157

DependentVariables(The Response)- Thresholds

AbsoluteDetection (absenceof noise) MaskedDetection(in presenceof noise) Flutter or FluctuationDetection

- Magnitude Loudness Noisiness

Althoughnoisinessand loudnessre listed asseparatedependentvariables forsubjective responseo impulsivesaunds_t will be shownthat they maybe takenasessentiallyidentical. However_asshownater in Section3.2.3_ theannoyanceresponseto impulsivesoundsmayt in somecases,be significantly different froma loudnessornoisinessmagnituderesponse.

Utilizing Iheabove frameworkof independentand dependentvariablesonloudnessor noisinessof impulsive(or fluctuating)sounds,an indexof the pertinentavailable literature is presentedin Table 2 which coversmostof the majorexperimentalstudiesof subjectiveresponseo impulsiveor fluctuating sounds. It will beconvenientto briefly review the pertinent findingsof theseexperimentalstudiesbythree generalgroupsaccordingto the type of stimulus.

PureTones Burstsof Noisee ComplexSoundsreal or simulatedimpulsivenoise including

heHcopierbladeslap)

3 - - 5


33/157

T ab le 2Index of ExperimentalStudies on Loudness/Noisinessof Impulsiveor FluctuaHngSoundsIndicating Experimental Variables Investigated

F.efereaces Sinqle ImT_flset(I) MulHoI_IRe-pentPdlImpL,ht.t (I)Narrow 6and W_c_aLand Narrow [_and W_de _aadNo Author Yw_r Tones ar Nolsa C_mplax N_e M_asur_d2) Ton_l _f NQI_n Com_le_ Noise Measured225 Hu_het (1946 0_F AT26 Garr_er &Millt (t947) D,F,L MT27 Mumon (1947) DtF, L LL28 Garner (1947) D,F e MT,AT29 Garner ((947) D*F AT,MT30 Miller (1948) 'D,L AT,LI.31 Garner (19,_) D,F,R,L LL32 Garn*r (1949) O_L LL33 Nietl (1956) D,F LL34 Green (1957) D MT35 HomIIton (1957) D MT36 i Pollock {1958) D LL37 _omp (1959) O,F MT3S M-Fodor (1960) D D LL39 Nits* ((960) D,L LL41 Sr_31[ _1962) D,L LL42 Port + (]965) D,L LL43 Sheilly + 0964) O MT44 Carter (1965) R LL45 Zw[cklr (1965) D D LL_MT46 Zipler 0965) 55 LL47 Garrllt (1965) D,R D,R LL48 Bcman 0966) D i LL50 Stewlnt + (1966) D LL$1 Zwlcker (1966) O LL_2 P*_ns (1967) AC, HBS N$a Peon_ (1967) L N54 Du_ov_kll + (1967) AM, R FT

dobson (1967) 5|,L LL,A_16 II_lulr 11967) O LL O LL


34/157

Table 2 (Concluded)S[n_F_mp_he_ Multlple Re.PeQledlmpulJe_

FPt_'r_n_ _'_O'rxgw Gand' Withe6and [ Notto_Band ' Wide fianceNo. Aulhor year To,_el of No_sl Complex No_le Measured Tone1 _ ef Nolte Complex No_ll Measured57 Horbert + (1968) R,_ _,RT FT58 Shepherd+ (19681 5B LL,A_9 Rothc;uJen (1966) R LL,A60 Johnson * I]969) 56 LL61 RelckQrdt + (]970) D LL62 Rdticha{dt (1970) D LL D kl.

6364 Fid_ll+ (]970) O D N,A D_,F D_R D_R N_A65 Ollefhfod (]971) L,AC LL66 ShlplQn " (1971) D D LL D dD LL67 Thom._o_ (t97t} D. D J.k D D LL70 Leverton (1972) H_ LL,A

72 Carter (1972) R_ET LL73 Stepher4 (19731 D D LL74 Carter (_9;_) R,_'T AT75 Boone (19731 D D LL76 I Leverton (]974) HB$ LL77 Pederlon (1974) D LL70 I Gustm'_on (1974) D,F,R, L LL D,F,R,L LL79 Tethatdt (19741 AM, A FTI]0 Fuller + (1975) D I A( ) Jndllpendlnt varlabhl idenllfied by abb_evtotedodl unde columnhead_ng 121 DependentvxrlabJ_m|a_uredJdentMed by obl:cevlatadca,de.

wM_h define type oF_lgnol_ p*arlt_l, narrowband_f no_le, comple__lgnol,I_depe_dentVo_abl_s Dependent Vcl]_blel

D- Duration SB . Sonic:Bo_mSignal A - AnnoyanceF - _lseF_equency H6$ . Hel_copfer61uda$1op AT - Ab_luteThrelkoldL - Signal /,evll AM . Amplhud# Modul_tIon FT - Flmler Th,'el_IdX - _epetlt]on _tl AC . Alrcroft._und LL . LoudnessevelRT- 9Jll(_ndOlcoy}Time _T- M_kadT_'_lhold

(*+ +col)


35/157

3.1.3 Subiective Responseo Impulsive Pure ToneSoundsFollowingpioneeringwork by Bekesyin 1929on the effectsof duration on25 26 27 28, 29, 31,32loudnessoF tanest Hughes, Garner and Miller, Munson, and Garner

lald the groundworkfor subsequentstudieson loudnessof slngle or multiple tonebursts.Typical resultsof thisearly workare representedby the data of Garner and Millert 26shownn Figure5. Figure 5a showsthe measuredslgnal-to-nolse ratio at detectionthresholdfor a single toneburstof varying durationpresentedin thepresenceoFawlde band maskingnoise. Figure 5b showsthesesame resultsnormallzedacco'rdlngtoa simpleempirical modelfor theauditory detection processcorrespondingto theoutputof a resistance-capacitance(RC)circuit. The latter isdriven by a signal (El) whichisassumedo representthe detectedenvelope of the tone burst. If weassumethat thetone is just detectedwhen thepeak output of the RC network reachessomefixed thresh-

' old detection level (Eo), then it can be shownthat for burstdurations(T), much lessthan the time constants "r= RC (analogousto the ear's time constant), the requiredsignal level increasesnverselyasthe pulseduratlon decreases, or

(RequiredSignalLevel, El) :" (r/T) (Detection Threshold,E ) (3)Thus, the productof thesignal magnitudeE1and the pulseduration T isa constant,as givenby

E1 " T_'E 1"= constant (4)Since theproduct of the slgnal magnitudeand pulseduration is a measureof the "energy"in the signal, thisrelatlonshlp is simplyanother way to define the so-called "constantenergy" law normally invoked to explain why_ for pure toneburstswith a short durationrelative to the ear's time-constant, the requlred slgnol level for detection increases3 dB forevery halving of the burstduration. Thisvery sameresult wasalsoobtainedby2 7Munson when a toneburstwasadjusted in level to equal the loudnessof a fixedduration reference tone longer thanabout 50 reset. However, as suggestedby Munson27and manyotherssubsequently,a simple "RC" circuit model Forthe ear'sresponsetotrans'entsoundshaso I re'tedappl*cohon.

3-8


36/157

40 ' i ' i b IIa)Measumd Dato O 400Hz

-_ D 67035 _ I000

0 1900Noise _eesufod inTelm*ofSp*cr_um.vol_ 30 _2- o

o 25

For


37/157

Thepractical implication of thismodelfor impulsiveno_seis that it couldoffera way toselect an optimumprocedurefor measurementf impulsivesoundsby dupli-cating, electronlcally, the ear's internal tlme-constant. Sucha rationale is thebasisfor the35 mstlme-constantselectedfor impulseprecisionsoundlevel meters(seeAppendixA). Unfortunately, there are severalco'replicationsn thissimplisticmodel whichare broughtout by the experimentaldata.

ReichardtandN_ese,6] employingo subjectpanelof 50 people, foundthat theloudnessmatchingof a toneburstof variable duration againsta fixed durationreferencetone, usuallyof the orderof 1 second long, wasa very difficult experimentaltask for theaveragesubject when the twoburst durationswere substantiallydifferent and led to ogreat deal of data scatter not indicated by the smallersubject panels(four to six) ia mostotherstudies. Byusingreference tonedurationsnear the middleof the rangeevaluatedfor the test tone, they foundmuch lessscatter in the loudnessbalances. On the basisoftheir refined technique, therefore, they measuredo time-constantof 30 milliseconds.61

Theserefinementsalso included a careful selectionof the temporalspacinganddurationof the testand referencesignalsto avoid possiblemaskingor memoryerrors incomparinga testand reference toneor toavoid what they termed"roughness"which wasobservedwhena rhythmically repeated pattern wasusedfor the testor referencesignal,particularly at pulserepetitionrates on the orderof 3 to 50 Hz (Reichard162). Thi._.squalitative measure, "roughness,"maybe importantin the evaluatlon of impuls;veaolse.

Other factorswhichcan causevariation in the observedtrade-off betweensignal level anddurationare: (I) the "energy law" failseither when Ihe signaldurationT is soshortthata substantialportionof Hsfrequencyspectrumalls outside thecriticalbandcenteredon the pulsefrequency(Garner29), or the signaldurar;onis muchlongerthan theear's time-constantr (2) the apparenttime-'constantncreasesas thesignal 26 75level approachesthe thresholdof bearing (Garner and Miller, Boone ), and (3)the time*constantapparentlyvarles with frequency,as ;mplledby thedatashownin

3-10


38/157

Figure5. It hasgenerally been accepted practlce, however_ to assumehat the tima-constantdoes not vary with frequency.

Thelack of agreement between investigatorson the time-constant still con-tinues. A recent study by Boone75an loudnessof repeated short tone burstsin noise,using20 subjects, produceda value For the tlme-constant of about 110ms. Terhardt79hassuggestedan RCtlme-constant of 13 msecto fit his unique measurementsof thedetection of periodic slnusoldal modulation (which he calls roughness)of pure tones.In summary, there Tssubstantial evidence to supportvalues for the time-constantranging from 13 msto over 200 ms (seeFigure 6). Becausethere is noapparent way toresolve this issueunequivocally for this report, the only practical choice appearsto be to work with the ex[stlng recommendationsor practice for the choice of time-constraints in impulseprecision soundlevel meters.

Thislack of agreement on the auditory t[me-constent is most unfortunate for it[mpliesIhe potential for conflicting evidence about a subjectlve correctio_ factor forimpulsive sounds. This point is illustrated in Figure 6 which showsthe potential rangeof the tlme-constant basedan the rangeof experimental data relating perceived loud-nessof an impulsivesoundversusits duration. Thus, for a given duration of an impulsivesound, Ihe potential increase in the signal level to achieve a loudnessequal to that ofa reference (nonlmpulslve)tone can he substantial. Clearly, any correction factorfor impulsive noisemustbe basedas muchas possibleon experimental date for subjec-tive responseo real impulslve sounds.

3 - 1 1


39/157

3_ 20 RengefMeasured /_ 15 tloncu Possible Range"5_o. _ of Loudness"5 _ 10 for 20 ms Pulse"_ u

.u_ 5"13 L._

m 0 I J ,I I I J f "t0.1 2 1 2 5 10 2 5 100 2 5 1000Pulse DuraHon, ms

Figure 6. Range of'Measured Sound Level - Duration TradeofF (where the Pulse is Judgedto be Equally Loud to the Reference Tone) Reported from Various Studies toIndlcate Possible Range of Uncertainty in Predicted Loudnessof 20 ms Pulse.(Note that the Time Constant Specified by I EC Sound Level Meier Specificationsfor "Fast"Would Tend to Fall Near the Middle of the Rangeof Measured Data --Adapted from P. Bruel, Reference 146.)

Pu!seRepetition RateAnother maj or varlable studied in loudness testsof repeated tone bursts is the

repetition rate. A very definiHve study in this area was n aported by Garner. 31 Fromthese results on repetitive tone bursts, a subjecHve correction factor for realimpulslve noises can be inferred His exper;rnental procedure consisted ofpresenHng, through monaural earphones to six subjeats, a oonHnualty repeated patternof a steady 1 second reference tone, 1/4 second silence, a I second cycle of repeatedtone bursts, a 1/4 second silence, ] second reference tone and so on. The repetitionrate of the repeated tone burst group was varied from 5 to 100 pubes per second, thepulse duration varied from 1 to 50 ms, the pulse frequency varied From125 to 8000 Hzand the intensity level of the pulse varied from 20 to I00 dB, The subject varied

3-12


40/157

the intensity of the tone burstsuntil he obtainedequal loudnessto the referencetone. Inmostcases, theenergy in each toneburstgroupwas lessthan thatof the equallyJoudsteadyreference tone. Thesubjectivecorrection A for equal loudnessfor thesetone burstsis simplysthe positive difference betweenthe soundexposurelevels of the reference endtest signals.Since the tone burstgroupandthe reference tone each lastsfor 1second, the difference insoundexposurelevels is alsothe difference in equivalent sound levels (L ). Thlssubjectivee qcorrection factor is shown_nFigure 7 for I000 Hz tone burstsand coversthe repetitionrate and pulsedurationrangeindicated. The intensity of the reference level was80 ,-lB.The typical variation of A with reference level and frequency is shownin Figure8.S

16 ' I ' ' IPulseDuration, ms1 _.80 dB12 -.0

_" 10, 820a-

..,oJI P

, o , I , , Ii. 5 10 20 50 100 200

PulseRepetition Rate, pulsesper secondFigure7. Subectlve Correction A for Repeated1000 Hz ToneBursts(fromGarner31 s

3 - 1 3


41/157

I I i Ia ) 1 0 00 H z

6

"D

0

-2 [ I I I20 40 60 80 100L (Ref), clBeq

1 0I l I I I

6

'< 4

2

0 r I J I I100 2 5 1000 2 5 10,000Frequency,Hz

Figure8. SubjectiveCorrection_ for RepeatedPureToneBursts,SPulseDuration20 ms, RepetitionRate = 25/Secondas aFunctionofa) ReferenceIntensity, Le (Ref)at 1000Hzandb) Frequencyfor the 80 dBReferenceLevel (FromGarner31)

3-14


42/157

As exhlblted in Figures7 and 8, thesubjective correctionfactor behaves in acomplex Fashion,even for simple tone bursts. TheseFiguresndicate the potentialdifficulty of developing any simple, general method for predlctlnga subjective correctionfactor Formorecomplex impulslve noiseswhichhave different spectra, ratesof attackor decay, or amplitudes. Nevertheless, Garner wasaerie to readily predict hls experi-mental resultsr suchas thoseillustrated in Figures7 and 8, on thebasisof two basicfactors:31

!. Thespreadingout of the Frequencyspeclrumof repeated tone bursts.As a result, the side band frequency componen_of the repetitivetone burst can fall into critical bandsoutside the one centered onthe honeburst carrier Frequency.

2. Theshapeof the.loudnessgrowth functlon. Due to thls unique shape,the loudnessof the s_deband componentsin eachof theseveral criticalbandsinvolved in this broaderspectrumof repeated tone Burstscan addup to o greater loudnessthan the sumof their energies because,at noiselevels well above threshold, the relative loudnessof e soundchangesmuchmoreslowly than the relative intensity (he., 2-to-1 changein loudnessfora 1g-to-1 (10 dB)change in intenslty). That Zs,the loudnessof soundssroughly proportlanal to the sumof the loudnessin critical bandsandthe sumof these loudnessvalues in the slde band componentswill not decreaseasrapldly at frequenciesremoved fromthe tone frequencyas the physicalenergles in theseside band frequencycomponentsof repeated tone bursts.

There isreally nothing new here, of course; it is simply the basic concept of loudnesssummationof complexsoundswhlch hasbeendeveloped into a fine art by Stevens,93,Zwicker, 89 and Niese. 86 However, appllcation of thesewell*developed conceptsfor loudnessof soundshashad only limited applicaHon to impulslvesounds.

It is important to recognize that the conceptof "startle" _snot involved in apredlction that a weaker impulsivesoundcan soundlouder than a strongersteady-statesound. This simplyresults from the accepted concepts for simulating the loudness

3-15

..... .+L ..... ....... .... , +.


43/157

perception of sounds. It remainsto be shownthat there may_ndeedbe on additionaleffect that makesimpulsive soundsmoreannoying than indicated by their loudness.

Frecluenc_Spectra of Repetitive ToneBurslsA brief consideration of the Frequencyspectra of repeated tone burstsis in order

here sincethis plays sucha primary role in the concept just outlined. As Figure 9 shows,repeated tone burstsproducea spectrum centered at the frequency of the tonewith s_debandsabove and below this frequency.

(T!It RepeatedTonegurst .-_1"J _'z-.,_-_._=12 1" -_

Figure9. Time Historyand Fourier SpectrumOfa Typical ImpulsiveSignalInAppendix (3, it is shownthat the inverseof the duty cycle of the pulse('r/T)

provideso qualitative indication of the number(N) of sideband harmonieswithin thenominal "1/2 power" spectralbandwidth. The more the repetition rate increase=,the _ouderthe pulsetrain will be, if the total energystaysconstant.31 With a very longduration, anda very slow repetition rate, all theenergy of the signal is concentratedina narrowrange of frequencies. IFthis rangefalls within a critical bandwidth, then theloudnessvariesas the signalenergywithin this band. On the other hand_ if this signalspectrumbandwidthis muchgreater than the crltlcal bandwidth, then loudnessof thesignalwill addapproximatelyas the loudnessof energy in eachcritical bandbutwith emphasison the loudestband.

Thebroadeningof the spectrumof toneburst=beyond the frequencyof the toneitself introducesan inherentcomplication in evaluatingsubjective responseo inter-mittent sounds. Thiscomplicationis overcome, in e senseFby usingo testsignal -broadbandrandomnoise, which already hase broad spectrum. Thus, the spectrumof

3-16


44/157

repeated burstsof wide band randomnoised_Ffersfromthe spectrumof the uninterruptednoise only at frequencies, which are generally infrasonic, correspondingo theburstrepetition rate. Spectra of s_ngleburstsof broadbandnoiseare nat significantlydiffer-ent from the spectrumof the steadynoise itself.3.1.4 Subjective ResponseoBurslsof Noise

Although Garner28 and Miller 30 carried out initial studieson responsetoburstsof broadbandnoise, Pollack36 presentedthefirstextensive studyutilizing pulsesof wide bandnoise. Forour purposes,hisresultsmaybe summarizedas showingthatthe difference between the L of a continuousnonlnterruptedbroadbandreferencenoisee qand the L of equally loud pulsesof the samenoiseincreasedfrom0 to+10 dBas thee qdutycycle of theburstsdecreasedfrom I Io 0. f. Forduly cycles belowO.1, thedifference remainedapproximatelyconstantat +10 dB. Thus, the subjectivecorrectionfactor A wouldbe +10 dBfor duty cycles lessthan0. I.S

Small, et al, 41 useda moreconventionalprocedureof balancingloudnessofrepeatedburstsof noiseof various durationsagainst interspersed1/2 secondburstsof aconstantlevel reference noise. They found that whenthesensationlevel of the referencenoiseburstwas60 dB(a typical listening level), the level of an equally loudvariabledurationtest burstwasconstantfor durationsdown to 15msecand then in=teasedby 12.5dB for each 10-to-1 decrease indurationForshortertestbursts. Forour purposes,thisis equivalent to the subjectivecorrectionAs increasinglinearly, at a rate of +3 dBper halvingof testburstduration, from a value of zero for 1/2 secondnoiseburststoa maximumof 15.2 dB for a 15msnoise burst, and thendecreasinglinearly at a rateof -0.75 dB per halving of testburstduration for shorterbursts. "1hisassumeshat I/2secondis the timebase for computingthe I of thevariable durationnoiseburst.eq

Garratt47 hasrepeated the testsof Pollack36 and Small, et al,41 with verysimilarresultsas shownin Figure 10where the measuredvaluesof ,', versusratio of test signal

Sduration to referencesignalduration isplotted. Similardata on loudnessof a shortburstof 2 to 4 kHz noiseFromBauer6 is also includedalongwith data on relative

3-17


45/157

noisinessof short bursts of noisebandsfrom Fidell ond Pearsons.64 The latter showlittleagreementwith the other datobut this may be due to the uniqueexperirnental tech-nique employed(flee-field presentation,interaction with a computarfor signal pre-sentatlon), and the "noisinessresponse"insteadof loudness. Thevalues for _ _n thisscase are actually differences in measuredA-weighted noTseevels of the reference andtestsignals. Theeffect of A-weighting on theshort noise burst levels is not clear.

20 I I I I 1 _ _ I I I d _3 _ 'Ref. D.fallon * NQ'leim_ Parlack 6 _" W}l_reS_all, _r ol4t 0.S _ Whito

T5 morn Garrott47 t.0 S_ V,l_ite"_ -- _ O Bo_er, et ol56 0.9 So_ 2-4 kHzFidorl 1.0 See Oclav0 &_' P_arso,_s4 * Wido_orDd

tO -- "Equol No;sleets1Annot,ancd._1_J (F_eo-Fiald)I

.jSl


46/157

0 to 45 dB. Figure 11 shows the resulting data obtained under one condition oF a repetitionrate oF I pulse per second (pps) and a burst duration of 1 ms. The ordinate defines theloudness level oF the composite signal relative to the loudnesslevel Ear continuous noise atthe same intensity as the noise peak. The dashed llne on the figure shows the computedL Forthis noise signal to illustrate, again, that the equally laud impulsive noise hasaneqL substantially lessthan the L oF a continuous signal with the same maximum level.eq eqThe resulting subiectlve difference Factor '_s approaches a maximum value of about 10 dBFora ratio of noise burst to backgrounJ nolse greater than 30 dB. Other data by Pollack 36and Garrett 47 on partlally interrupted noise gave similar results as shown in Figure 11.

0Pollack (Figure 1 oF Ref. 36)

I "_ Pollack (Figure 4 of Ref. 36)-5 l- ",_". o Pollack (Figure 5 oF ReF. 36)I "_ x Garrett (Ref. 47)Drawn to F_t Data

"_ -1 0

_ -15 _\__ /.- Equal LoudnessLevel, '< o-20 ', ,

J- o-25 ,_ A _ 10 5dB-30 Computed L .............eq-35 I I I I I t10 20 30 40 50 60 70

Burst to Background Ratio, dBFigure 11. Difference Between the L oF the Repetitive NoTseBurst Superimposed on aeqSteady Background Nolse and the Level oF a Contlnuous Noise which Sounds

Equally as Loud as a Function of the Ratio of the Burst Level to the Back-ground Level (Burst Duration : 1 ms, Repetition Rote : 1 pps)

3-19


47/157

In summary,with theexception of the resultsof Fidell and Pearsons,4 theexperimental data on loudnessor noisinessof shortburstsof randomnoiseshawcon-slstont trendssimilar to the toneburstdata in termsof order of magnitudevalues forthe subjective difference Factor. Limited resultson the'ear's time-constant fromthesestudiesare nat inconsistentwith the values observedFromthe pure tone testS. AdditionalsupportFor the time constantvaluesdiscussedn Section 3.1.3 is providedby data by Dubrovskiland Tamarklna54 on"subjectiveperceptionof the relative loudnessof amplltude-madu/otednoise. They hypothesizea tlme-constantfar theear of 10 msto explain their data -a value slmlfar to the 13 mscited earlier for testson the modulationthresholdof puretones.

l_e appllcatlon of a "tlme-constant" modelagain appearsconvenient to explainexperimental results. However, thisdevice mayindeedbe misleadingbasedon theunique resultsand resultlng hypothesisposedby Miller 30 in his studyof the delay indetectability of a low level noisesignal fallowingthe interruption of a higher levelmaskingno_se. Basedon his results, Miller suggestshat:

'_... the auditory systemasa whole doesnot have a fixed rate of decayof sa many decibelsper second independentof intensity. _us theauditory systemcannotbe said to have a "time constant" in the sensethatthis term is generall); used,and we havebeencareful tousethe term"critical duration" instead. Thisis not tosay that the mechanismof theear hasno time constant, however. As in air mechanicalsystemshereisa finite timerequiredfor the ossicularchainand the cochlear fluidsto beginand to stop their motions. Themechanical time constantsofthis system_however, are far toosmalltoaccountfor the 65 msecperiodsof perceptualgrowthand decay.

'It hasoften beenconvenient to liken theauditory systemto an integratingcircuit ..... Theevidence seemstoshowthat the ear is notsOmuchanintegrating device as it isa delaying device ..... According to ourhypothesis, the growthof the perception of na_seis the integralof the

3 - 20


48/157

.J

distribution of transm_sslon_mesof the various pathways from the cochleato the higher center, and not the integral of the soundintensity. " 30

3, 1.5 LoudnessVersusNoisiness of Impulsive NoiseNone of the preceding studies cited on noise burstsemployedstandardloud-

nesscalculation proceduresto predict thek results. However1 bath Pollack36 and4/ . ..Garrett usedd_fferent ernp_rJcalapproachesbasedon we_ghtlngtheFrno_seburstsignalsby a Function related to the observed level-duratlon trade-off. Garrett wasparticularly successful_npred_cHngthe loudnessof 48complex transient signalsconsisting of repeated decaying s_nusolds. Other examplesof this approachare coveredin Section 3.1.6 on responseo complex impulsive sounds. However_before consldeHngmore complex impulsive sounds, let usexamine one final (and very recent) study on theloudnessand noisiness of no_sebursts.

A new and uniqueapproach to the predlctlon of humanresponseto _mpuls_venoise is provided by thework of Izuml. 82 in order to examine the passiblesubjectivedifference between loudnessand noisiness,IzumJconducteda setof two laboratorycontrolled psychophyslcolexperiments. In the Firstexperimenta periodically inter-mltlent pink noisesignal wasused todetermine if therewas indeedany differencebetween these two subiectlve parameters. Uponfindinga significant difference, thesecondexperimentwasconductedsothat an effective assessmentmethodcouldbeestablished.

For the first experiment, consistingof two phases,subjects wereasked to om-pare_ using thepairedcornpaHsonmethod, the test signal (intermittent pink noise)witha standardsignal(continuouspink noiseat 70 dBA). "111eigna lswere presented_nafade-in, Fade-outsequenceas shownin Figure 12. In order to overcomeany passlbleerror due to sequenceblast thestimuli were presentedboth signal f irst andstandardIirst for an equal numberof times. Thewhole procedure wasrepeated for six differentburst-tlme fracHons(Bl1:); the BTFbeing defined as the signal-on time divided by theon-plus-off time.

3-21


49/157

[ I I Ill I JI I0 3 5 t31e,_[* t? 2S_G 30 (_ec)

Figure 12. Time-AmplitudeSequenceDiagramof the StimulusPresentation(fromIzumi82 )During PhaseT, the subjectscomparedthe palr of signalsn termsof their

relative loudness,i.e., howmuch louder (or softer) thanthe continuoussignal is theintermittent signal? During Phasel] the samesignalswere replayed, but th_stime thesubjectscomparedthemin termsof their relative noisiness.

The resultsof bath phaseswere tabulatedandcomparedwith each other (seeFigure13). From theseresults, Izumi concludedthat ".... asfar asperiodicallyintermittentsoundsare concerned, laudnessudgmentsand noTslnessudgmentsaresigniflcantly andsystematicallydifferent. Therefore, Ioudnessandnoislnessshal]baconsideredas different attributes."

Once he determinedthatthe two parametersare indeeddifferent, izumi .-atup hissecondexperiment in order toarrive at a mode[which wouldaccurately predictthenoisinessof an intermittentsignal.

in thisexperimentthe subjectswere presentedwith signalswith 25 differentBTFs. Theywere asEedafter each trial torate the relative noisinessf the signalsas in the Firstexperiment.

3-22


50/157

f r f-- LEFT COL.:"LOUDNESS"

R IGHT C OL. :" N OISINE SS"

lo-- _ _N5--,,"_m ' --

_ --

16/250 6311000 6301100031/500 250/1000 95011000TIME PATTERN

Figure 13. Result_of Experiment]. (_ki=a (Loudness)Data and Yahomashisa(Noisiness)Dataare ComparativelyPlotted. Filled CirclesRepresentMean Relative BurstLevelsJudgedby EachSubject.Averages andStandardDeviations are Shownby Cenh'ol Linesand Rectangleson BothSides (from lzumi 82)Mode lFromthe resufh_of thosetriafs, Izumi developedwhot hecalls the "Perceived

NoisinessModeJof Periodically Intermittent Sounds75-A."

LRB=61OgloBTF+(IOIogIoRR+ 10)(1 -e "15TfF) , dB (,5)

3 -23p__ .... ......... ._ .......... .,, .... ........ . ...... .


51/157

whereLRB= relative A-welghted noise level of burst in dBBTF= burst time fractlon_ i.e., on-tlme/on + off timeRR= repetition rate per second

Totf = off time in secondsIn order to test this formufar hepredicted the value of LRBfor the25 intermfftent

noisesusedin Experiment I]. The LR_'swere colculoted using nine different methods: peakburst levels in terms of LoudnessLevel_ StevensLL(S); LoudnessLevelt Zwlcker LL(Z); PerceivedNoise Levelt PNL; and A-weighted noiselevel; A-weighted equivalent soundlevel; Pollock's36

47method;Garrett s method noise rating number (NRN) as specified by ISO1 la and Model 75-A,proposedby izumi. 82

The predicted levelswere thencomparedwith the experlmentoldata. The resultsare shownin Figure 14.* From theseresults, lzumi_sModel 75-A appearstobe the bestpredictor. The other methodsalways underestimatetheperceived noisinessof the inter-mittent sounds,

Startle EffectThe majorreason,according to Izuml, for the differencebetweenloudnessand

noisinessis the startle effect createdby the intermittence of the sound. Thestartleeffectis basedon three physicalparametersof the signal: repetition rate, rise time and theburst-to-backgroundratio.

In theseexperimentsthe risetimeand the burst-to-backgroundatio wereheldconstantandonly the repetition rate wasvaried. Thecontributlon of repetitionrate to thenolslness-loudnessifference wasq_ntified end thisinformotion_shownn Figure 13e_wasusedin the developmentof Model 75-A. Accordingto lzumi_ work is still necessaryf the

contribution of the startle effect is to be understood.

*Figure 14 is a corrected versionoF the form publishedin Reference82 which weskindly suppliedby Dr. lzuml.

3-24


52/157

It will be pointedout later that onestudyof subjectiveresponseo helicopterblade slapaso functionofrate ofslop23 hasalsoshowna trendoFincreasingapparentnoisinesswith increasingrepetition rate although the range of "pulse rate" exploredwaswell abovethat (i.e., 10 to 30 pps)exploredby'IzumL

,,, i I I I , I I I I I I I I rO" NOISINESS"I,, LOUDr4ESS"i=

.J

.Jj,_ 5m

i, Jp-w

O O._lllfl I f i , _ I_2 4 8REPETITIONRATE, PULSESPERSECOND

Figure 13a. Comparisonof Loudnessnd Noisinessvers_, PapefitionPatefor BurstTime Froctmnof 0.063 (from [zumP'_),

3-25


53/157

10

rn'13c

O_-J

oD -5--"1t

o -1 0 --_ JI- -Qt_e en. -15 -

-20 '

iFigure 14. PreliminaryValidation of A=essment/v_tho_s. Errorsof Predictionare Calculated for 25 Intarmlttant Noises in Experimentt. MoanErrorsore Shownby Central Linesand StandardDeviationsbyReatangleson BothSides(fromlzumi82).In summary,althoughthis is only onestudy, Izumi showsquite well that

nots;heSSnd loudnessre nat the samesubjectivequantitieswhendealingwith inter-rn|ttontsounds,and that the startle effect of the intarrn_ttantsounds a primecauseof this difference.

3-26


54/157

3.1.6 SubjectiveResponseoComplexImpulsiveSoundsEarly work on subjective responseo morecomplex impulsivesoundsother thantoneor noiseburstsinvolved rneasuringloudnessof short h'iangular transientssuchas44 72repeated gunblasts. For example, Carter, Carter, andCarter and Dunlop74

explored the loudnessand thresholdlevelsof this typeof transient, which had a pulseduration of I ms, Farvarying rise times(. 05 to 0.5 ms)and repetition rates (1 to256 pps). Theeffect of repetition rate wasadequately coveredby a simpleenergy rule(+3 dBincrease in _ntensltyto ma_ntalnloudnessForeach halvingof duration). Forthe highest repetition rate, the ratio of on-time tooff-time neverexceeded0.S and wastypically much less. As with all the preceding _mpulsivenoisestudiescited so Far(except Fidell and Pearsons),earphonepresentationwasused. The loudnessudgmentswere madeby comparisonof a 3 secondreference (white noise)signalwith two impulsesseparatedby 1secondfrom each otherand from thereferencenoise. For mostof theloudnesstests, the reference noisewas Fixedat 15dB abovethreshold (sensatZonevelof 15 dB) Foreach subject.

Theprlnc|peJresult fromCarter's work is i_e evaluation of alternate meansofpredicting loudnessof triangular impulsivesounds. For eachrepetition rate andrisetime, the loudnessof the referencenoiseand impulse_at lf,e '*equally feud" intensitylove/s, was catculated from the signal spectra. Thespectrawere computedFromthe pressuretime history for the irnpu/slvesoundsand measureddirect/y for thereferencenoises.

Thefour calculation methodsanalyzed were: Zwlcker87, 89 Stevens, Mark V193 Perceived Noise Level103 A-WelghtlngTheaverage difference between the calculated loudnessbasedor, the com-

putedspectrum)of the ;mpulslve sound,which was judged equally as loudas the refer-ence sound, and the calculated _oudnessf"the reference soundwasmeasuredor all the

3-27


55/157

combinationsof rise timeandrepetitionrates(306 cases). The resultsare summarizedin Figure 15for thesefour methodsn termsof thisdifference asa functionof pulser e p e t i t i o n r a t e .

,+1o ._10i i i i _ i i i lie ] I _ I _ I I I I I(c) lJ t It - LR,

d_ -I I ! I I,.--_L r I f I/ -i I I I I I I f I I

" o o ;_ x x.-__ x_ x _ _ xoO X X X X- 'i I I I I I r 1 ._ I T I i I I f ! IIS _ 256 l 4 16 SS 256

RepetitionRate, pulsesper secondFigure 15. Comparisonof Loudnessalculation Methodsfor TriangularTransients.Variation aboutIheoverall mean(within each loudnessmethod)of the

meandifference(aver suhiects)between the calculated Iou:lnessLt)of the triangularI msimpulseand thecalculated loudnessLR)of thereference noisessubjectively judgedto be equQIly loud.Deviation fromzerois a direct measureof the error in each loudnesscalculation method=o) Zwicker, phons;(b) StevensMarkVI, phons;(a) PerceivedNoiseLevel; (d)A-Weighted Level. The symbolsdenotevaryingrise tlme (e, 0.5 ms;+, 0.25 ms;o, O.1 ms;andx, 0.05 ms). (FromCarter72)

Surprisingly, the loudnessomputedon the basisof the A-weighted levelsexhibits the least deviationaboutan overall mean. TheZwicker methodwasnext inaccurecy. There is reasontodoubtthe generalapplicability of theseresults, how-ever, asshall be seenwhenIheseoudnesscalculat ion methodsare applied to othertypesof impulsivesounds.

Fidell and Pearsons3 investigatedthe influenceof phaseof harmoniccom-ponentson thejudgednoisinessf five different simpletransientsoundscorrespondingto (1) an ideal N wave, (2) anN wave wlth 1 msriseand decay times, (3) a triangularwaveform, (4)a squarewaveform,and (5)a doubletor podtlve and negative sharp

3 - 2 8


56/157

impulse. Power spectra Foreach basic waveform were maintained essenHally constantwhile phase was adjusted by a computerized waveForm generater. No significantinfluence of phase on subjectlve loudness was detected.

They also evaluated the subjective loudness of 12 actual impulsive soundsandeight artificial sounds presented, as were all their signals, over a high quality loud-speaker system. The characteristics of these 12 soundsare listed on Table 3. Thedifference between a tlme-lntegrated objective measure of the sounds and the samemeasure for the reference sound is shown in Figure 16.

Table 3Description of Naturally Occurring Impulslve SoundsEmployed as 64Comparison Signals in Evaluation Experiment by Fidell and Pearsons

Duration ApproximateImpulse (m_ec) IdentTficaHon Spectral Charaaterlatlcl

I 300 Aulornahlfa Door Slam Peaks at 0.5 kHz2 150 Paper Tearing Near flat spectrum to TOkHz3 425 Hand CIo_ Rhes and foils about 0.8 kHz4 450 Two Bottles CIInkTn_ Highly leptokurHc at 4 kHzToge the r5 580 Chain Col/alvin _ on Itself ! Near flat spectrum to I kHz, foilsslowly at higher frequencies6 4B0 Noclurnai Animal Noise CompJex spectrum peaked at 0.125and 2.5 kHz7 180 Squeaky Release of Air peaks at 0.g, 1.6 and $ kHz

Through a Valve8 400 Balloon Burstln_ Peaks at 0.2 kHz9 600 Balloon Burllln_ Peal_ at 0.2 kHz

tO 180 Automobile Horn Discrete frequency peaks cancentrafedbetween 0.3 and ] kHzI I 1200 Simulated SonicBoom Predominantly Jew frequency, falling

i_eaply from O. 125 kHz12 900 Basketball Bounce in Highly Energ concentrated between 0.2 endBaverberanl Envlronrnant 1.6 k_'lz

Standard 1000 While Noise, I Slo/_d Octavo Band from 0.6 to 1.2 kHz

3-29

:_,_._......


57/157

-S i I L IO8a0 O0

O0 0 D 0aa Ooo _[] __jD_O__

+5 - ---_D'O--- 8Oo- ___o_9__ 8oa o0 03 +_0 8o_ o n oO, 8o -_-----,oo ; o_o_ 0013+15 0 0 0 Naturally Occurring impulses[] Synthetic Impulses

0 A Synthefic impulse+20 with LowFrequencyEnergyMean of Data

A

+25 I I I ILeq(C) Leq(A) Leq(N) EPNL

Figure 16. Comparisonof Time-Integrated Measuresof the Impulsive Noises with the SameMeasureof tho EquallyNoisy ReferenceSound(1 Sec of Octave Band Noise, '600-1200 Hz). (From Fidell and Pearsons4)

b


58/157

The ordlnate specifiesthedifference between the overage soundlevel (com-puted Froma meansquareaverageof the drgltTzed trine history of the signal)64 andthe samemeasurefor the equally noisy impulsive sounds. Forthe A-welghted measure,thTsdifference is identlcal to our subjective correction factor _s and wasequal to 12.5dBwith a standarddeviation of 3.5 dB. The standarddeviations for the other measureswere s)ightly greater, thus indicating the A-weighted average soundlevel wassl_ght)ymore reliable asa predictor ofnoislness of these impulsivesounds. Note that theseimpulsive soundsvary substantlafly }n their characteristics; somemaynot be very impul-sive. However, they are all essenP[aJlyingle events and not repetitive. The averagevalue at"6sobserved, ]n this case, hasconsiderably more validlty than the valuesg_ven up to now Forthe fo)low]ng reasons:

T. It wasmeasuredw_tha loudspeakerpresentationthus rnsurlngthatrealistle head diffraction effects are included.

2. The objective measurementof the average sound level shouldbevery accurate - they wereperformed by dfgltar analysisof"arecording of the actual soundreproduction.

3. The soundscover a variety of actual impulsive noisesto whichthe subjectscan relate.

4. The instructionsto the subjectsasked for a judged nols_nessutpromptedon annoyanceresponseaswell (i.e., the test instructionsdefined a noisy soundasannoying, unacceptable, objectionable,and disturbing if heard in the home during the day and nlght). 64

Loudnessmeasurementsof decayingslnusoldal transientssimilar to those usedby Gorrett were carried out by Gustafsson8 but at soundlevels from95 to 117 dB.While the resultstend to substantlatethoseg_ven earlier_ the high noiselevels usedplace thesedata outside the area of interest for this study.3.1.6.1 ]SO RoundRobinTests

Themostcompleteset of data onloudnessof impulsivenoisesis provided bythefinal resultsof an international cooperativeRoundRobintest programorganized

3 - 3 1

'11 '


59/157

under the auspicesof the International StandardsOrganization, ISO/TC 43/SC-1,StudyGroup B, "Loudnessof Impulsive Sounds." The f_nal rel?ort, preparedby Pedersen,et al_77 representsresults from 22 laboratories and "close to 400 subjects. " Add_-tionel detailed supporting data were reported by Shipton, Evans, and Robinson, fromthe National Physical Laboratory,66on the specific results from their testswith theISO RoundRobindata tapes. Detailed informaHonan findlngs of the ISO RoundRobinTests, drawn from these two sources, is presentedin Appendix B.

Although the testsconsistedof an evaluationof subjectiveand objective cor-rection factors for the following three typesof impulsivesounds,resultsfor only thesubjective correction factors for the first groupare consideredhere.

Group I Nine quasi-steady impulsivenoisesecorded fromach_al sourcessuchasa teletype, pneumatichammer,outboard motor.

Group If Five single impulsenoises,suchas froma gunormechanicalram.

Group [I1 Six 1kHz tone pulsesof 5 to 160 msduration.Thesoundswere presented tothe subjectsvie budspeokerin repeatedA-B sequencesandmatched, in loudness,with referenceslgnalspresentedat three soundlevels (55, 751 and95 dBre 20 /sPa). Theoverall grandaveragesubjective correction factorl "s' for allreportinglaboratories, nearly 400 subjects,and for the nine repetitive noisesin Group bis 12.5 dB. Thestandarddeviation over the nineaverage valuesfor each noise is 0.9dB. Thisisa highly smoothedstatistical resultsincethevariation betweensubjectsforany onelevel andtest soundcan be l0 to 15dB. However, it isestimatedthat thefinal result is reliable within :kl.5 dB. No estlrnate couldbe madeof subjectivecorrectionFactorsfor the five :.ingle impulsesoundssince the equivalentnoiselevelsfor thesesoundswere not available.

3 - 3 2.... =.--


60/157

3.1.6.2 Loudnessof SonicBoomsTheevaluation of the loudnessof sonicboomsprovidesadditional information

pertinent to the subjective responseto impulsivesounds.Zep/er andHare|46 success-fully predictedthe relative loudnessof sonic boomsoundsbyapplyinga loudnessfrequencyweighting to the Fourierenergyspectrumof thesimulatedN waves. Johnsonand Robfnson5' 60 have extended this type of approachto successfullycorrelate theannoyanceesponsefromexplosiveblastsand sonicbandsaswell as conventional air-craft saundson the sameloudnessscale. Theyutilized the S.S. Stevens,/V_rk V|,/oudnessalculation method93 wlth a modification to extend fts low Frequencyangeto encompasshe strong, very low frequencyenergy inherent in sonicbooms. This lowfrequencydeficiency in theloudnesscalcu/ation methodshasbeen observedby others.64' 67Howevert thismaynorbea significant problemForthe type of impulsivesourcesof concernin thisreport.

A key element in Johnsonand Robinson'spproachis theuseof a specific 70 msintegrationtime for measuringthe signalspectrum, Thiswasintendedtoduplicate theear's integrationHme.60 Note thatthis is twice thevalue of"therlme-constantspecifiedfor the impulseprecisionsoundlevel meter. This isobviouslya critical pointthat will require careful considerationin the selectionof an optimumimpulsivenoisemonitoringtechnique.

Johnsonand Robinsonpplied a loudnesscalculation schemeto the predictionof annoyancefor impulsivesources. Are these two formsof humanresponse(loudnessand annoyance)really synonymous?The answer, basedonavailable data is that theyare not nocessorHythe same. Thfspoint isfundamental to describing impulsivenolseand deserveshe morecareful review taken up in the next section.3.2 Annoyanceand Other Sub]ectlveResponseso |repulsive Noise

Reviewof theexisting literature dealing withannoyancedue to impulsivenoiseyields a wide range of approachesand result. Theseresultsfrom available studies,

3-33


61/157

excluding thoseonhelicopter bladeslap, are briefly summarizedin Table 4. Annoyanceof helicopter blade slap is consideredlater. Annoyancedue to aircraft sonicboomswere ofprimary concernin about haft of thestudiescited in Table4. While moststudiesattemptedto measureannoyance, the terms"unpleasantness'1 and "unacceptabilily"12 were alsoused. |t wasassumedhat thesetermsrepresenteda similar measureof subjectiveresponse. The qualitative descriptor"annoyance" is notwell defined butmaybeassumedo representan overall subjective reactionto an impulsivenoisestimulus.Thisreaction mayvery wen integratenat only the loudnessor noisinesssensationbut alsothe responseo other non-acoustlc factorssuchasstartle, emotionalcontentor intrusivenoiselevel relative to the ex_stingbackgroundambient level.3.2.1 AnnoyanceResponseo ImpulsiveNoise

A divisionof the referencesonannoyanceresponsesnto the three categoriesof impulsivenoisestudiesdefined earlier helped in selectingonlythoseapplicable tothiseffort. Category I, which is the principalconcernof this report, coversthe"repetitive impulses"producedby two-strokemotorcycles,rock drills, pavementbreakers, helicopter blade slap, and otherrepetitive impulsivenoisesources.Category|[, "single impulse,"includessonicboomsandartillery blasts. Category I|l,"unsteadynoise," coverstraffic and subsonicaircraft noiseand isactually mareconcernedwith noise "events"rather thanwith "impulses." In onesense_however,the Firstand last categoriesore similar, differing basically in the time scaleofand

between "events" and in the crest Factoror ratio of maximumpeak pressureo rmspressure

In thestudiescited, correctionFactorswere developed to accountForannoy-ance on thebasisofone or morefeaturesof the impulsivesounds:numberor Frequencyof impulses,amplitude, fluctuation (rate of change in amplitude), and durotlon. Someinvestigatorsproposedcorrection factorswhich were applicable to impulsivenoiseingeneral, independentof ffs characteristics.11, 1la, 1lb Thus, EldredI1 and ISOR199611aproposea 5 dBcorrectionshouldhe added to any communitynoisewhich is

3 - 3 4


62/157

Table 4A SummaryoF L_teratureonAnnoyance Responseso Impulslve Noise (Excluding Studiesfor Helicopter Blade Slap)

Nalse Saur_o Param_eers Mealurernenfs Retult !NolSe NO, el Irnp_Jfle bType Authol Date Subject1 lemt ReFere_l_e Vazled CoFis_anr SubieCl]_e C)hje_l_Ve Ba_e Scare Ca root;on ( Relpont_ S_ule

PJ_lchlk L 1957 4 Tone I_ulst tqc*r_ Freq_rlc_, D,Jrotlon Ju*t NaDaeabl_* Atle_ua_axlRopetiti_e Repetltloa Urlpteotanrnel* _dfiPulses Rare, dB

Kelghle_ .8 1970 1903 Live OfFice No_ Acceptability Peak Index (PIJ Average ! 3.S3 P[ ]/2 AcnplubllltyAvera0e d_. LA I d8 c

Anderson, 10 1971 24 Recorded None Durofionond Bk0d L_vel Adject6.e Pair L L 14a, dBd) AnnoyanceBobinwn Rood Drift Nun'ber of E_p,_ure ( I I Paint eq eqBollts Duration Compaei_oa)

EJdred f9961h: ]97t Community None Souece. Lecel Community teq Leq 5dE CommunllyNaheSO'R Na_leS Sire CompJoln_D Sq_.Iwlo_t Leve_

Fucfm13 1972 100 Ra_o_d_d Tone Bursl Hondclop SPL I ttan,_lap Anr_,_y_nce LA ' LA AnnoyanceHondcrnp Tone B_,_sl Dula t;onDuraHon

If Broadbent, 2 19L.4 79 Jab PropA,'C ;Jel Level Durallon Anr_yan_ LpN LpN "{e) Anr_yancaSingle Robinson _;onla Beam Source RaringB_tlky 3 1_96s i 3_(} 0 LI_I None O vetpeellure A nl:oyonce O _'erplelsure_nJc Boom fpsl) ProDabUIty

o1 l_f

Kry_ 7 1970 Sonic Boom Sublo_ia Jet U_aep_ab_l_t_ LpN EPNL 2_ (I Unacce_hablJlty7-_ (_ LPN)' dS

1973 iArlilfer)'_ )_o_0 Anr_yancn CNR EPNL t0 Jog]0_, dB Communityahor_r I& 5_,ra_e _rost Complalnt_ R#_1) on_lSonic Boom

CHABA 2] 19_5 AililJery, A n,x_y'_nca Led n Lc 0 for Community !lie"WG _69 Blasi_, C_wel_h_ed Sl3on_a& Structural_ni_ l_oom l_veJl Vibrm_on


63/157

Table4 (Concluded)

No_ Saurce Pa+ameterl p_,etal.tom+ nh Rnlultsi_lo_le No. oF mpu .e br_'pe Aulhor Date Subjects lntt Reference Varied Cnnl_nnl S_Jbiecfive Obincr_e Bate Scnle Cmrect_on Regpanle Scale

IJl i_b[nlon 6, 9 J 1969 [raffle. / 1/_e;ory L 2.5_ a, d EI Ann_ anceUm_dy Aitcrofr eqNo iN

PartY,i 2 J972 A ffcmfr 13_eory PNL 0 (I_._mrion) Acce plab;lilyPa.y

Fullet, 15 1973 24 Traffic None LA(rnaN) Event Annoyance LNp Lq 2.56 a. d8 Anr_y_neaJt_b[n_n Dumtlon,Fmquency

)2 dt (g) Gl.lmlMallchot, 14 1973 352 Live Norm Level, Tolemblllty, L, L , NNI, Lq j.(dLt Sub]ectlv_lullJr, A ]l_m |t J[_JrOlon_ and Ac ti'_ily eqZtmmirman Frequency of ail!ur_:Qr_e, I'Np Reaction IQ[yen h A nno_,ance i_o_le(o) e,ated on c_nceptt by Rmenblll_ a,_d St_+e_ ,_ ,'ited In R+lfe_enoe I I.(b) [_ept at noltd, I_lU_] to tub]ectlve Correction /aclae _ in dJ), to Cmtect lot lu_Jct]w responle Io irnf_tslv_ rmiw.() P[, Ihe hmk ;n_l_, II Ihe _umof nurd bero [ Impeller which are 5, 10, 15und 20 _ _bo_ Leq ;n I mlnu/_ _r_ple.(d) _ - Tllmpo_ Sl_ndgrd _eviel[on.(e) _)n[c b_em, wlfh LC - I 1 6 dD ;n(_oon equaJly annay[ng al )e t n oi_ _td_'t with LpN _ 110 PNdfi.(1) _ LpN DIIferer_ce in LpN, In PNdB, ben_een Impul_ive soundond bockgtound r_lle, rltpectlvely.(g) Comlcllon propaetionol to mean _lU_re mt_ o; level Flucluolion,


64/157

deemed impulsive, while iSO R19'991]b recommendsa Fixed10 dBcorr:tion be app)iedtoassesshe hearingdamagerisk of impulsivenoise,

in mostcases,as noted in Table4, the varioussubjective correctionFactorsdevelopedFromthe referencedstudiesare added to the measuredL of an impulsiveeqnoise in order toobtain the effectfve L of a nonimpulslvenoise thatwill producee qthe sameannoyance. Forone method, however, the correction is not tobe addedtoL but rather to the EFfectivePerceived Noise Level (EPNL). 7 It mightbe assumede qthat thiscorrectionfactorproposedby Krytor could alsObe applied to the L scale.,,,21 e,qAnothercorrection method(developed by CHABA Working Group o7) rove yes the predic-tion of annoyancethat may, in partt be dueto building vibration inducedby impulsivesounds.Since this responses potentially quite important for large amplitude impulsi_.,eounds

and since it is not treatedin any of the other studies, it is also includedin the tableHowever, as clearly pointedout by the CHABAWorkingGroup, the concept, basedon the useof"a C-weighted L without Furthercorrection, was designedto be appli-eqcabte on)y to single )_igh_ntensityimpu)si

epa noise paper

Documents