1,

DOCUMENT RESUME.

ED 125 848 SE 018 715

AUTHOR Ward, W. Dixon, Ed. 44,

TILE Proceedings of the International Congress on Noise asa Public Health Problem (Dubrovnik, Yugoslavia, May13-18, 1973),

INSTITUTION Environmental Protection Agency, Washington, D.C.Office Of Noise Abatement and Control.

SPONS A-GENCY .American Speech and Hearing AssOciation, Washington,DC.; Environmental Protection Agency, Washington, D.C..; Union of Medical Societies of Yugoslavia,Belgrade.; World Health Organization, Geneva(Switzerland) .

PUB DATE 73'

NOTE 29p.; Not available in hard copydue to marginallegibility of original dopument

EDRS PRICEDESCRIPTORS

IDENTUIERS

ABSTRACT

MF-$0.0-Plus P,00stage. H6 Not Available from EDRS.*AcouStital Environment; *Conference Reports;Environment; - ;*Environmental Research; *HealthEd7Ucation;- Physical Health; *Pollution; PublicItealth;"Reports; SpeeChes*NOise'pollution

This paper is from the proceedings of a secondinternational conference on noise as a"public health hazard. Funded,by the Office of Noise Abatement and Control of the EnvironmentalProtection Agency, these conference proceedings serve as a sourcematerial summarizing all known criteria that could be used inestablishing, national standards for noise contr61.,. The conferencelimited content to the effects of noise on health, thus eliminatingdiscussions of a technical engineering or, legal nature. Numerous,topics discussed at the conference include: noise-,,iiduced hearingloss;-community response; and sleep diS,turbance dile to noise. Thepaper presented here concerns ,noise aid its effect on the hearing ofspeech. In this research, tht author reports that the speechinterferepce level depends upon a number-of varying factors. Amongt,ese are the octaves chosen for speech'communicationr, the types ofspeech discrimination/ tests used, and the wearing of ear plugs ormuffS by notmal and hearing loss subjects. (MA)

***********************************************************************Documents acquired by ERIC include many informal unpublished *

* materials not available from other sources. .ERIC makes every effort ** to obtain the bit copy available. Nevertheless, items of marginal ** reproducibility are often encountered and this affects the quality ** of the microfiche and, hArdcopy reproductions ERIC makes available ** via the ERIC Document Reproduct1dn Service (EDRS). EDRS is not* responsi4e for the qualitrof the original document. Reproductions ** supplied, by EDRS are the best that can be made from the original.,04:***********************************************4********w*********

LP41.

U S OEPARTMENT OF HEALTH,EOUCATiON t WELFARENATIONAL INSTITUTEOF

EOUCAT ION

THIS oocum Nor HAS BEEN REPRO -DU, ED xAC EL Y AS RECEtVE0 FROMTHE PFRSONORORGANICATIONORIGIN-4TiN( -.IT POIN fSt)F VIEW OR OPINIONSSTAIN") 00 NOT NECESSARILY REPRE-SENT OF ,AL NATIONAL INSTITUTE OFE

AT:ON POSITION OP POLICY

550/9-73-008IF:

,x\-c E. ST.4)..

0 AMY

1114 p846i.e.P

ROCEEDINGS of theINTERNATIONAL CONGRESS

. on NOISE'as aPUBLIC HEALTH PROBLEM

kr)

. ,JDUBROVNIK,,, YUGOSLAVIA-,IVIay 13-43, 1913

ty

1,

U.S. ENVIROI4MENTAL,PROTECTION AGENCYWasin ,20460

AQ)

PROCEEDINGS of theINTERNATIONAL CONGRESS

on NOISE as aPUBLIC HEALTH PROBLEM ,

DUBROVNIK, YUGOSLAVIAMay 13-18, 1973

\ft,

a

SPONSORS of the CONGRESS

Union of Medical Societies of YugoslaviaEnvironmental Protection Agency, U.S. Gov't

American Speech and Hearing AssociationWorld Health Organization

Prepared byTHE U.S. ENVIRONMENTAL PROTECTI N AGENCY

Office of Noise Abatement and ntrol3 W. Dixfn Warc(, Editor

S

Foreword

.0 . f------ ..

In 1968, a Conference on Noise as a Py lic Health Hazard was organized by theAmerican SpeeCh and Hearing Association. t this conference, an attempt was made tobring together a grOup of speakers who co d present summaries of the current state ofknowledge on all aspects of the "noise pro em", ranging all the way from fairly technicaltreatises to completely non-technical stat ments of personal opinion. Such a wide-rangingrepresentation'was judged to be necessa for the purpose of thatconference, which was topresent a broad overview of what "nois pollution" was all about, to government personneland other intelligent laymen who saw p at it was probably going to become a hot issue, andgive at least alfew examples of the sci ntific evidence underlying arguments about just'whateffects noise does have. i

At that time it was realized/ that as the environmentalist movement gatheredmomentum, a rapid development of/public concern could be expected, and so a permanent .

Committee of ASHA was established, one of whose charges was to plan another conference'when it was judged appropriate.

The burgeoning of interest in' noise in the intervening 5 years has clearly met, if notsurpassed, our expectations at that time. In the developed areas of the world, millions ofdollars or their equivalent are being spent on surveys of noise levels and exposures, andincreasingly stringent noise regulations are being imposed by all levels of government. And,although the measurement of t e effects of noise is nowhere near as simple as themeasurement of the noises thems Ives, many laboratories, mostly with federal support, areengaged in full -time research on he hearing losses, sleep disturbance, speech interference,alteration of physiological state, a d annoyance caused by. noise. g

Accordingly; in 1971, we b gan looking for a sponsor for a ssecond conferenceonewho would agree, we hoped, to fund Attendance by a substantial number of researchersfrom abroad, so that certain are of knowledge less intensively studied in the USA could beincluded in the subject mattef. ortunately, the head of the newly-created Office of NoiseAbatement and Control (ONA ) of the Environmental Protection Agency, Dr. Alvin F.Meyer, had need of just such a onference, as a source material for a document summarizingall known criteria that might bp used to establish national standards for noise controlthatis. provided that' the Congress ;Massed the bill, then being duly debated and amended, thatwould make such a document/necessary. Furthermore, certain PL 480 funds (money thatmust be spent in other coLintries) were available, which meant that the degree ofparticipation by foreign scien: sts might be even greater than we had hoped. Not only that,but the particular PL 480 fu' ds in this case were in Jugoslavia, the country that includesone of the garden spots of the' world, Dubrovnik.

On the assumption tha our Congress would pass some ,form df the bill in question(which it did on October. 2; , 1972), we forged ahead with plans- for our meeting, nowupgraded to an International Congress. With the help of.Dr:Grujica 2arkbvia, the energeticPresident of the Jugoslavian' Medical Association, and Dr. Maris) Levi of the University ofSarajevo, a planning meeting was held o which we invited a representative from most of thecountries in which noise rest arch was eing done (i say "most" because we could not quiteafford to pay for attendees rrom Japa ;Australia, and South Africa because of the distance'involved, even though consi, arable res rch is being done there). At this meeting the formalagenda was decided'on, ancl the list of nvited participants prepared. It was agreed that wewould try to limit the Con ress conten strictly to the effects of noise on health, thereby

excluding discussions of engineering aspects of noise reduction and control, descriptions ofmethods 'for legal control, and presentation of viewpoints of Special- interest groups. Therewas some debate about how much time to allot to public opinion surveys of annoyance,some of us contending that annoyance, as measured in that manner, iS not a health hazard atall in the ordinary sense of the term. However, proponents of the WHO definition of"health'4, in which any deviation from "optimum well-being" is regarded as undesirable,carried the field, and the final day of the Congress was therefore 'given- over to thesociologists.

Despite a series of crises precipitated by governmental red tape originating both inWashington and Belgrade, the Congress"was held on May 13-18, 1973 at the Libertas Hotel

' in Dubrovnik, We had two rnajol' disappointments; one was the failure of our Russianinvitees to appear due to the fact that our official invitations had not been sent early.enough. The other was that the Xerox machine at the Libertas was out of commission.However, the general success of the Congress can be gauged by the fact that the audiencewas as large on the final afternoon as at any other tare.

A side benefit of,the Congress (or so we hope) was the forination of an internationalorganization consisting of 5 "tears " who will try to accumulate and coordinate knowledgeabout the effects, of noise on (1) ternpordry and permanent hearing loss; (2) extra-auditoryfunction; (3) .speech; (4) sleep; and (5) community reaction. The parent group, or "bask"team, will attempt to'cons.olidate this knowledge for use by governmental agencies, and willmake plans for the next Congress. Although the organization is,now alive, its name is still inquestion. At the moment it is still the "International Scientific Noise Teams", but theresulting acronym has a negative connotation that pleases few of us. Other names are beingconsidered.

I regret that the length of the i vited papers made it impracticable to publish at this-time any of the short contributed p ers that were presented at the Congress, many ofwhich were excellent, or the often-liv ygliscussions that followed each session. It is hopedthat these cl.n be included if another rinting of the Proceedings is to be made.

An enterprise of this scope can of be a success without hard work on the part of manypeople. Without doubt the most effort of all was put forth by Dr. Levi, who managed all themechanical details of the -Congress, with the help of his and Dr. 2arkovies staff,particularly, FeliA Vesna.' Of Dr. 'Meyer's staff, David Bach deserves special thanks forhandling the__

oft-complicated travel arrangements of the participants.Official thanks are extended to our sponsoring *olanizations: The Jugoslavian Medical

Association, The American Speech and Hearing Association, the World Health Oiganization,and of course most of all the. Office of Noise Abatement and Control.

Finally, I would like to thank my fellow participants (with two exceptions) for gettingtheir manuscripts to me rapidly so that, with the help of Miles Kahn and Jean Pellegrini ofONAC and quick work by the Government Printing Office, the Proceedings of This Congressare appearing before they are out of date.

W. Dixon Warddi"tor

27 August 1973

I

a

SPONSORS OF THE CONGRESS

Union of Medical Societies of YugoslaviaWorld Health OrganizationEnvironmental Protection Agency, U.S. GovernmentAmerican Speech and Hearing Association

.----

CO-PRESIDWTS OF THE CONGRESS

Grujica 2arkovrC, M.D. President of the Union of Medical Societies of Yugoslavia;

Alvin F. Meyer, Jr.

Robert Goldstein,

/Deputy Assistant Administrator of the U.S. EnvironmentalProtection Agency

/Angelova, ,BqjgarivBorsky,P./USACurlee, USACerkez, ., YugoslaviaDieroff, H. DRDiumhUr YugoslaviaGulian,E., maniavan Hattum, , USAHinchcliffe UKJansen, G., GermanyJokic-, J., ugoslaviaKonig, E., witzerlandKryter, K. USA

President of the American Speech and Hearing Association

PROGRAM COMMITTEE

Kylin, B., SwedenLevi, M., YugoslaviaStanding SecretaryNixon, C., USAPearsons, K., USAPinter, I., HungaryRaber, A., AustriaSpoor, A., NetherlandsSulkowski, W., PolandThiessen, J.G., CanadaTobias, J.i USAW. D. Wald, USAPresidentWebster, J., USAWhitcomb, M. USA

A

TABLE OF CONTENTS

Sessionil

Introduction and Masking Effects of NoiseChairman: Henning von Gierke (USA) Pag)

A Preview of the Congress Content, W. Dixon Ward (USA) 3Systems of Noise Measurement, Karl S. Pearsons (USA)

_

The Effects of Noise on the Hearing of Speech, John C. Webster (USA)Reception of Distorted. Speech, Jerry V. Tobias, F. Michael Irons (USA)Hearing Loss and Speech Intelligibility in Noise, Jerzy J. Kuiniarz (Poland) . 57The Long-Term Planning of a Noise Contr61 Program, Michael J. Suess

(Denmark) 73

Session 2

Noise-Induced Hearing Loss (NIHL)Empirical DataChairmap: D. Robinson (UK)

Basis for Percent Risk Table, Aram Glorig, William L. Baughn (USA) . 79A Critique of Some Procedures for. Evaluating Damage Risk from Exposure to

Noise, Karl D. Kryter (USA) 103The Incidence of Impaired Hearing in Relation to Years of Exposuie and

Continuous Sound Level, (Preliminary Analysis of 26,179 Cases), A. Raber(Austria) 115

Some Epidemiological Data on Noise-Induced Hearing Loss in POland, ItsProphylaxis and Diagnosis, Wieslaw Sulkowski (Poland) 139

On .the Problem of Industrial Noise and Some Hearing .Losses in Certainc, Professional Groups Exposed to Noise, J. Moskov (ffilgaria) 157

loise-Induced Hearing Loss_from Exposure to Intermitant and Varying Noise,_

W. Passchier-Vermeer (Netherlands) 169Evaluation of the Hearing-Damage Risk from Intermittent Noise According to

the ISO Recommendations, B. Johansion, B. Kylin, S. Reopstorff (Sweden) 201

Noise-Induced Hearing Loss from Impulse Noise: Present Status, R.R.A. Coles,C.G. Rice, A.M. Martin (UK) 11

Hearing Loss Due to Impulse 'Noise. A Fieisl Study, Tadeusz Ceypek, krzy 4Kuiniarz, Adam Lipowczan (Poland) 2/ 219

Hearing Damage Caused by Very Short, High-Intensity Impulse Noise, H.GDieroff (DDR)

IV

229

Session 3

Noise-Induced Hearing Loss7MechanismChain is H.G. Dieroff (I1DR), R.ilinchcliffe (UK) Page

-1/4-

Behavioral, PAological and Anatomical Studies of Threshold Shifts inAnimals, Donald H. Eldredge, James D. Miller, John H. Mills, Barbara A,Bohne (USA) 237

Presbyacusis in Relation to Noise-Induced Hearing Loss, A. Spoor(Netherlands) 257Noise Exposure, Atherosclerosis and Accelerated Presbyacusis, Z. Bochenek, W.

Bochenek (Poland) 267High-Frequency Hearing and Noise Exposure, John L. Fletcher, (USA) 271Susceptibility to TI'S and PTS, W. Dixon Ward (USA) 281Growth of TTS and Course of RecoVery for Different Noises; Implications for

Growth of PTS, Wolfgang Kraak (DDR) 293Experiments on Anima 1g Subjecf to Acute Acoustic Trauma, Wiktor JankoWski

(Poland) 301

Session 4S

Interaction of Noise with Other Noxious Agents in Production, of Hearing LossChairman: E. Lehnhardt (BRD)

Influences of Chemical Agents on Hearing Loss,.M. Haider (Austria) 307Hearing Loss of Forest WOrkers and of Tractor Operators, (Interaction of Noise ,

with Vibration), Istvri Pinter (Hungary) 315Infrasound and Hekring, Charles W. Nixon, Daniel L. Johnson (USA) 329The Effects of Airborne Ultrasound and Noir Ultrasound, W.I. Acton (UK) . . 349

Session 4-Br

Performance and BehaviorChairman: D. E. Broadbent (UK)

Psychological ConseqUences of Exposure to Noise, Facts, and Explanations,'Edith Gulian (Romania) 363

Similar and Opposing,Effects.of Noise on 'Performance, L. Hartley (UK) . . . 379The Effects of Different Types of Acoustic Stimulation on Performance, C

Stanley Harris (LISA) 389Behavioral Effects and'Aftereffects of Noise, David C. Glass, Jerome E. Singer

(USA) 409Effects of Noise on a Serial Art-Term Memory Process, G. Wittersiicim, P

Salame (France) 417The Effect of Annoying Noise on Somer.PsYchological Functions During Work,

Irena Franszczuk (Poland)

V

425

Vas

Session 5

Non-Auditory Physiological and Pathological ReactionsChairmen: E. Grandjean (Switzerland), S.,Kubik (Czechoslovakia) Page

Non-Auditory Effects of .Noise, Physiological and Psychological Reactions inMan, Gerd Jansen (Germany) _.I

. , 431 'Industrial Noise and Medical, Absence, and Accident Record Data on Exposed

Workers, Alexander Cohen (USA) 441 ,

Factors Increasing and Decreasing the Effects of Noise; D.E. Broadbent (UK) . . 455Examples of Noise-Induced R actions of Autonomic Nervous System 'during

., . Normal Ovarian Cycle, Bar ara,Griefahn (DDR)1 59

The Influence of N ise on A tory Evoked Potentials, J. Gruberov, S. Kubik,,

_J. Zaleik(Cz hoslovakia ., 469Some Data on th Influence of Noise on Neurohumoral Substances in Tissues

,and Body Fluids, Lech Markiewicz (Poland) 473Stress and Disease in Response to Exposure to NoiseA Review, Gosta

Carlestam, Claes-Goran Karlsson, Lennart Levi (Sweden) 479Sorge Laboratory Tests of Heart Rate and Blood Volume in Noise, Karl D

Kryter (LISA) 487

Sess 6E16

Sleep and Its Disturbance by NoiseChairmen: B. Metz (France), Levi (Yugos

EffeCts of Noise on SleepA Reyiew, Harold Williams (USA) 501Predicting the Response to Noise During Sleep, Jerome S. Lukas (HA) 513The Effects of Noise-Disturbed Sleep on Subsequent,Performance, M. Herbert,

R.T. Wilkinson (UK) 527Effects on Sleep of Hourly Presentations of Simulated Sonic Booms (5 N/m2),

William E. Collins, P.F. lampietro (USA) . 541'Prolonged Exposure to Noise As Sleep Problem, Lav\erne C. Johnson, Rich,ard

E. Townsend, Paul Naitoh, (USA),'Alain G. Muzet "(France) 559Relationship L etween Subjective and PhysiologiCal Assessments of

NoiseDisturbed Sleep,. A. Muzet, J.P. Schieber, N. Olivier-Martiii, J.Ehrhart, B. Metz (France) 5-75

the Effects of Aircraft Noise on Sleep Electrophysiology as Recorded in theHome, Gordon plobus, Joyce Friedmann, Harry Cohen,, Karl S. Pearsons,Sanford Fidell 587

Noise and Mental He'althAn Overview; W. Hausman4USA) 593Observations of the Effects of Aircraft Noise' Neir Heathrow Airport on Mental

Health, C.F. Herridge, L. Low-Beer (UK) 599

V i

*),"04,

Session 7

Community Response IChairmen: G. Thiessen (Canada), P.N. Borsky (USA)

Methodological Aspects orStudies of Community Response. to Noise, Er landJonsson, Ola Arvisllsson, Kenneth Berglund, Anders Kajland (Sweden) . . 611

Decision Criteria Based on Spatio-Temporal,Comparisons.of Surveyspn AircraftNoise, Ariel Alexandre (OECD) 619

PsychO-SoLialwFactors in Aircraft Noise Annoyance, Aubrey McKennell (UK) . 627A Survey op Aircraft Noise 'in Switzerland', Etienne Grandjean, Peter Graf,

Anselm Lauber, Hans'Peter Meier, Richard Muller, (Switzerland) 645Aircraft Noise Determinants foil the Extent of Annoyanceteactions, Ragnar

Rylander, Stefan Sorensen.(Sweden) 661Reaction Patterns in Annoyance Response to °Aircraft Noise, Stefan Sorensen,

Kenneth Berglund, Ragria Ryland& (Sweden) 669The Reduction of Aircraft Noise Impact Through a D7namic Preferencial

R6nway'System, Martin Gach (USA)

Page't

679A Causal Model for Relating_ Noise Exposure, Psychosocial Variables and

Aircraft Noise Annoyaned, Skipton Leonard, Paul N. Borsky (USA) 691Community Responses to Aircraft Noise in Large and Small Cities in the USA,

Harrold P. Patterson, William K. Connor (USA) . 707. -

Colighlunity Response IfCliirman:- R. Rylarider (Sweden)

Session 8I

,

Measurements of Street Noise' in Warsaw and Evaluation of Its Effect on theAcoustic Climate of Dwellings, Schools, Offices, Hospitals, Hotels and

- Parks; the Degree': of 'Offensiveness to Inhabitants in *the,, Light of aQuestionnaire, Alelcsander Brodniewicz (Poland) - 721

A New Field Survey-Laboratory Methodology for Studying Human Response toNoise, Paul N. Borsky, H. Skipton Leonard (USA) 743

An Interdisciplinary Stu on the Effects, of Aircraft Noise on Man, BRohrmann, R Schumer, A. Schiimer-Kohrs, R. tuski, Finke(Gerrany) 43

Rating the Total oise Environment. Ideal or Pragmatic Approach? D. W:Robinson (UK)

765

777Motor Vehicle Noise: Ideneification and Analysis of Situations Contributing to

Annoyance, William J. Galloway, Glenn Jos (USA)

V I ,

1 0

785

I

40.

Session 9

Summary and IntegrationChairmen: G. -iarkoviC (Yugoslavia), W.D. Ward (USA) . Page

Summary, I.J. Hirsh (USA) 807

.of

1i

4

0

it

4/

.4

o.

THE EFFECTS OF NOISE ON THE,HEAFIING OF SPEECH,-

'Jo lm C. WebsterI-

Naval Electronics Laboratory CenterSan, Diego, California 92152 '

To coyek, this subject matter, i will talk about the Xrticulation Index or AI, and theSpeech Interference Etvel, or SIL,-and in particular the relatioh betWeen them. I will, show

:that the best octaves to choose in calculating the SIL depend on what Artidulation(Al) you want to work at or design for. And to male tis meaningful, I will have to showyou what scores you can expect to get 'on syllable, word, or sentence tests at variousArticulation Indices. Beyond this,, I will discuss what sort of tests can be used to testsystems or listeners operating at high AIs, that, is, in relatively twet environments. If thiS'seems off the subject, I will relate thest typed of tests to methoddZivaluating hearing aidsand/or different kinds ofhearing losses inclUdingnoise-induced ilearing losses.

Tct talk of these 'things intelligently, I will have 'to, spend a little it of time disdussingThe pros and col of efficient intelligihility tests. At the first of these conferences (Webster,1969)1 traced the-earlyhistory of intelligibility testing. I will not repeat it here, but I wouldlike to stress a single distinction made by the earn' Bell Telephone Laboratory investigators,namely, articulation tesfinkas opposed to intelligibility testing. Articulation testing involvesthe use of nonsense syllables to eter, ine what single stech sounds, phoneme,s, distinctive/I-features; or consoikants are mash . Once any aspect of redundancy or language,entersthe testing it is no' longer articulation but intelligibility that is being tested. Articulationtesting centers on speech sounds per se. Intelligibility testing involve both, the ear and thebrain or involves both speech sounds and language.

To summarize very briefly the problems associated with speech tedling; I must mentionthat the construction of speech intelligibility tests varies along two dimensionsthe redun-,dancy and/or vocabulary size of the inpulktimulus (language) and the constraints or numberof 'possible choices in, the output or response. Wjthin'ocabularies of the same' size therelative familiarity of the word and the number of syllables in-a word and the contextwithip which it is imbedded influence its intelligibility. The constrahtvn the respoQse,open vs. closed sets, also affect intelligibility scores. Closed set or Modifiect4Rhyme Tests(MRT) -(House et al., 1965; Clarke, 1965; Kreul et al), 1968) and pseudb-clOsed set rhymeword tests (Fairbanks, 1958) are largely repladircg,the ctpek-set Phonetically Balanced-(PB)(Egan 1948) and Spondee tests and other multiple choice_ tests -at the present time. Themajor reason is the. time and effort required to thin both talkers and particularly listenerS in

-7the open-sA PB-type word test.This is not the document to trace out in any more detail the history, the rationale, the

strengths and weaknesses, nor the actuallistirigs of syllables, wOrds,,yhrases,'or sentencesused this century to evaluate the effects of noise on speakers, listeners, coqimunicationcomponents and"systems, etc: Recently, 'however, Webstez (1972) hai compiled 2,4 lists of .

word,. phrase, and sentence tests in English. A very good reference for more details onintelligibility tests is Clarke, Nixon, and Stuntz (1965): because ii has abstracts of,,over 1.60earlier references.

)r.

Table 1 shows the relationship between intelligibility scores and Articulation Index (aspecial form of speech-to-noise rati ) of many standard speech' tests. The generalizations tobe made from Table I are that the er the stimulus vocabulary and/or the size of theresponse set, or the more redundant in ter, s of context, the higher the score for a given Alor speeckty-noise ratio.

Table 1

Expected Word or Sentence Scores for Various Articulation Inices (AI),

4

-

AI PB* MRToc*, SENT*

0.2 22 54 77

0.3

t'1

72 92

0.35 50 78 95

0.60

0.80

85

9Z

94 ,

98

98

99ca,

*From Kryter and' Whitman (1963)-

**From Webster and Allen 972)

ti

So far I have mentioned ony the printed stimulus and response variables that affect'intelligibility testing. The talkers, listeners, and the noise environment around them havevery large effects on test validity and reliability. For example, Dreher and O'Neill- (1957)had 15 naive speakers read in 5 different noise levels.-When the words and sentences,wereplayed to listeners at a constant speech-to-noise differential the speech originally recorded, iiinoise was the more intelligible. Pickett (1956) shows, however, that if vocal effort measuredone eter in front of the lips exceeds 78 dB, intelligibility drops.

,,,,should be apparent by now that intelligibility test results require some interpre-t is neither simple n.or straightforward to assess the affects of noise on speech using

Word t ting methods. It would be advantageous to specify the effects of noise on speech'interms p the spectra and level of the noise and.oc; the speech. Two such physibal schemes

26

13

exist ,the Articulation Index, AI, and the Speech Interference Level, SIL. The ArticulationIndex or Al assumes that there are 20 bands in the speech spectra betw'een 200 and 6100 Hz'

'that differ'in bandwidth such that each band contributes 1/?.0 of the total articulation. Eachband contributes linearly to the extent that the speech pek level exceeds the RMS noiselevel by from s,..0 to 30 113,., The Al is a specialized method of specifying the speech-to:noiseratio. It is a non-dimensional.numeric that varies from zero to one, but it can be consideredto be,,a decibel scale ranting from Zero to 30 such that for example an Al of 0.5 corresponds

.to acomplex signal -to -noise ratio of 15 dB, 0.8 to 24 dB, etc. -:,-

he Al Was- introduced' by French and Steinberg (1947), gels-eralized and simplified by4B anek (1947a), and refined .and validated by Kryter(1962,a, 1962b). The Al was discussedat the first Congress of N9i'se as a,Public Health Hazard by Webster (1969) and by Flanaganand Levitt (1969) in suffkient detail that it will not belabored further here. .

- Almost simulOneously with the introduction oVthe Al, Beranek (1947b) proposed asimplified substit to for it, the Speech Interference Level (SIL) of noise:The defi,nition of),/SIL is the ari metic average of the decibel levels in three or four selected. Octaves. Thechoice of octaves will be discussed later. The SIL is only a measure 'of noise, and to interpretit in terms of permissible distances betWeen talker(s) and listeIner(s), reference -mast be made,to a table' (Beranek.(1947b)) or a graph (Botsford (1969), Webster (1969)). An updating,Mark II, of the Webster (1969) graph is shown as Figure ,I. It.differs frdm Mark I unveiled atthe first of these conferences by (I) adding two new physical Measures, the four .octave

. PSIL (.5/1/2/4) and the proposed SI-6(Lweighting which will be disaissed in more .61etaillater; (2) appending an AI scat" to help orient people in the real meaning of the figure; and(3) a droopoff in the communicating voice level curve to reflect the fact that at voice levelsabove 78 dB intelligibility does not increase as fastowith vocal effort as at lesser levers. Thegist of the figure is that for an Al of 0.5 using "normal" vocal effort (65 dB at 1 meter)conversation at 16,feet or 5 meters can Jake place in noises tiigh as 50 dB as measured onthe A-weighting network of a sound level meter. .

The one aspect.of Al that has been alluded to by many (see Webster (1965)) but notfully appreciated is tha t as the Al-and its cizrrelate, word intelligibility, increase, the,mostirpportant speech frequencies and/or the frequency range of noise that masks the speechmost 'electively increases *from between 800 Ad 1000 Hz to between 1700 and 1900 Hz.This of course should be reflected in the octaves chosen to calculate the SIL, and thisrelationship will be developed in the next four figures. .

Figure 2 shows a method- of calculating the Al by counting the proportion of dotsbetween the noise spectra and the upper limit of the conversational level speech spectrum.The example shows how it can be used to specify the Al for a.-6413 per octave (-3 when

nmeasured in octaves) noise.This figure was developed from the Cavanaugh et al., (1962) procedure of deriving Al's

from dOt patterns spaced in a 30-dB range in the shape of the normal male speech spectrum:The concentration of dots reflects the relative importance of different frequency bands tothe intelligibility of speech, heard in noise. Figures 3, 4, and 5 show the results of caleitlatingAl's at 0.2, 0.5, and "0.8 for 5 theorelical noises, and show why and how the octaves chosenfOr SILs should vary accordingly. ,_,- Note from Figure 3 that the spectra lines cross each other (with about a 2 dB spi'ead)at l000 Hz. Since these are all well- behaved; theoretical noises with constant,slopes, the

br

27

14

S I

-60

SIL

(LLB

)35 30

5

40

23 ,F

EB

197

3

5565

7585

95.

50.6

070

8090

32A

PP

RO

X A

l FO

R N

OR

MA

L V

OIC

E,

1:1.

10.

80.

60.

40.

2

8 2

A L

EV

EL

30.

SIL

(4)

21

PS

IL 2

3

,NO

ISE

-DIS

TA

NC

E A

RE

AW

HE

RE

FA

CE

-TO

-FA

CE

CO

MM

UN

ICA

TIO

NS

INN

OR

MA

L V

OIC

E IS

PO

SS

IBLE

AI >

0.5

NO

ISE

-DIS

TA

NC

E A

RE

AW

HE

RE

UN

AID

ED

FA

CE

-TO

-FA

CE

CO

MM

UN

ICA

TIO

NS

AR

EA

IMP

OS

SIB

LE

4N-

4,t,

94,A

.c(

se,,

"'-'7

N' e

o/4

,46.

6tis

,

<9,

5Nq.

1., /

-...,

l e

4050

3141

3343

6070

51.

61

5363

\el

8 0,25

8090

100

110

7181

91'

101

7383

g310

3

Fig

ure

1. N

eces

sary

- vo

ice

leve

ls a

s lim

ited

by -

ambi

ent n

oise

for

sele

cted

dis

tanC

es b

etw

een

talk

er a

ndlis

tene

r fo

r sa

tisfa

ctor

y fa

ce-t

o-fa

ce c

omm

unic

atio

n. A

long

the

absc

issa

are

var

ious

mea

sure

s of

noi

se, a

long

the-

ordi

nate

dis

tanc

e, a

nd th

e pa

ram

eter

s ar

e vo

ice

leve

l. A

t lev

els

abov

e 50

dB

(A)

peop

le r

aise

thei

r vo

ice

leve

l as

show

n by

the

"exp

ecte

d" li

ne if

com

num

ieet

ions

are

not

vita

l or

by th

e "c

omm

unic

atin

g" li

ne if

com

mun

icat

ions

are

vita

l. B

elow

and

to th

e le

ft of

the

"nor

mal

" vo

ice

line

com

mun

icat

ions

are

at a

nA

I1

leve

l of O

S, 9

8% s

ente

nce

inte

lligi

bilit

y,sh

out,

com

mun

icat

ions

are

pos

sibl

e ex

cept

abo

ve a

nd to

the

right

of t

he "

impo

ssib

le"

area

line

,

0

90

cr

a3 80Ocr

"0 '7000'0ta 60r03

50

40

ta.;

< 30

0

20

OCTAVE, PASS BANOS IN CYCLES PER SECO 740:20.45 90 - 180 - 355 - 710 - 14 00 2400 - 5800 - 1 1 200

ml,.....,weishiligAligh

hilbialsibLinallMilMili1;111PRIII)

m---v44.74:7114-14m

65 ,

93 "N111111111Pth

%44411

9 .1000 2000 4000 8000..

u .

-.gym

."'"ft-A,Lit7711-""41114

3 125 250

.500

5 2 5 2 5100 1000

4FREQUENCY IN CYCLES PER SECOND

10000'

Figure 2. Al speech region for "conversational level" speech. The number of dots in each band signifies therelative contribution of speech in that band "to the Al. A series,of idealized thermal noises with -6 dB /octspectra are drawn in 5 dB steps. The number of dots above each noise contour is proportional-to the Al ofconversational level sped' in that level of noise. (After C.s...aliaugh Pt al., 1962.

.19

16.

f

I I 0

cc.

In I00O

" 9000'Oci

Or80

co

zTO

ww_1

Z 60cO

< 50

40

'45OCTAVE PASS BANDS IN YCLE3 PER SECOND

90 160 355 770 14 00 21100 5600 11300

r

o ,lig

. 1\

I

i

+6

.

1"...

\\ //

.

.

.

if

70.

FLAT

17. #

..

se/ ..

.

.

.

.. o,

\A

At x O. 2\

,

N\\\\

A 63

.

1 5. , 25 500

.

\'41000 2000

\\ N.,,,,,,\ r)\

4Q00 N .

9..

. Q... I7,1 1 I 1 ^7 I 7 1 I T r

5 I 2 5

tt 100 1000FREQUENCY IN CYCLES PER SECOND

5fr

1000

Figure 3. AlloViable octave band sound pressure levels of steady state noises with spectruM slopes of -12, -9,-6, flat, iand +6 dB per octave for an Al of 0.2 and conversational level speech. The superimposed SI-70contotir is a proposed frequency weighting network for evaluating the speech interfering aspects of noise atAl = 0.2.

30

1r

crossing point at 1000 Hz s also the SIL for he octaves centered at 500, 1000,and 2000Hz (.5/1/2 SIL). Note also that thespectra--cr ss a hypothetical line at 1414 Hz (.5/1/2/4SIL) with a spread of abo t 9 dB and that th: spread at 2000 Hz (1/2/4 SIL) is about 18dB. Obviously, the .5/1/2 IL is the measure with the least variability for. specifying thelevel of diverse-spectrum noises at an Al of 0.P, which corresponds roughly to Fairbanks(1958) Rhyme Test (FRT) and Modified Rhym. Test (MRT) score of just over 50%, a 1000word phonetically Balance (PB) word score of ust under 25%, and a sentence score of justover 75%.

Interpreting Figure 4 iM the same way, it is avident that (1) at'1000 I-r(.5/ SIL) the--spread is about 10 dB; (2) at 1414 Hz (.5/1/2/4 IL) the spread is minimal, about 2 dB; and(3) at 2000 Hz (1/2/4 SIL) the spread, is about dB. It is equally apparent therefore thatthe .5/1/2/4 SIL shows the least variability in secifying an AI of 0.5 which corresponds toa PB score just over 75%, an MRT (and FRT score of about 90%, and a near' perfectsentence score.

Figure 5 shows the 1/2/4 SIL to be the least variable in specifying an Al of 0.8 whichresults in,near-perfect socres on all word and sent nce testing materials.

It, should now be apparent that the choic- of octaves in calculating SIL is directlyrelated to' the intelligibility required of the systep to be evaluated or to the AI expected ofthe system. But just to summarize it once more le us look at figure 6.

NOte for example -that the slope of Al vers SIL decreases with decreasing SIL levels ,as the spectral slope changes from -12 to -6, to to +6 dB per octave. It therefore follows,and it is evident from 6 that when these theoretical noises are equated in le41 togive arrapproXimlitely equal Al of-0.1, as measur' d by the .5/1/2 SIL, they are not equal af-Als of 0.5 or 0.8. Wlien equated by the .5/1/2/4 SIL,, the noises are generally equivalent inlevel at an AI of 0.5 but not at 0.2 nor 0.8. Finally, if equated by the 1/2/4 SIL they aregenerally equivalent at an AI of 0.8 and not at 0.5 not= 0.2.

. ,interpolation shows that_ a 50% PB score (Al = 0.35) could be about equally: well

specified, over a 'large diverse sample of noises by an ..5/1/2 or a .5/1/2/4 SIL. An. AI of0.35, MRT (FRT) score of 80 and sentence .score of 95% has been recommended as theminimum .acceptable specification for certain military communication equipments' (seeWebster 'and Allen, 1972) operating in highly adverse environments. Even lower levels foracceptance have been suggested for use in past (see Webster, 1965) and thus lendredence to using the .5/1/2 SIL for measuri g the effects of Navy noises. Architects and

others working in quieter environments and requiring higher levels of communicationefficiency naturally prefer. Als- of 0.5 for Which the .5/1/2/4 SIL is the least variablemeasure. Only the perfectionist would need to design or operate at Al levels of 0.8 and sothere is probably no serious reason for considering the 1/2/4 SIL for practical engineers.

Probably the best valihating data concerning the' hange of SIL frequency with Al arethose of Cluff (1969). Cluff equated the spectra did lev s of 112 industrial noises to giveone-third octave AIs of 0.1, 0.2---0.9, and then determin the bandwidth, that gave thebest prediction (least standard diviation) over all noises foi (1 n average level in one thirdoctave bandsf-shnilar to an Sit ---(2) an overall or band level similar to a C-weighted(but band-limited) sound level meter reading---as well as broadb measures of (3) theA-weighting, and (4) the proposed SI-70 weighting. He found as the X' increased from 0.1to 0.9 the center frequency of the optimum bandwidths increased fr m 848 to 2264

31

16

110

or

col00cr

cs.1 90- 0*0

00.

Ili '80cr

ZJ

707

">.

Cs 'SOtZa)

U(t 80

0

40

i.OCTAVE PASS DAND3 IN CYCLES PEN SECOND

49 10 ISo ]9D 710 14 00 --- 2 800 e400 11200

S1-6 Al = 0,5

-1'2

\ ..

.

e SIN

\\ Cr-->7

_

.

.

-6 `S.\\\\ .

,tI..e

sr ,

. .r ,

N,

3FLAT

129 2 50 100,

dr/

,'1000

\

2000

\\Av. 000 000

,

1I 2 5 1 2

100 1000FREQUENCY IN CYCLES PER SECOND,

10000

Figure 4. Allowable octave band sound pressure levels of steady state noises with spectrum es of -12, -9,

-6, flat, and +6 dB per octave for an Al of 0.5 and conversational level speech. The superim SI-60contour is a proposed frequency weighting netWoric for evaluating the speech interfering apsects of no at

Al = 0.5.

32

JO

IF-00

40

20

45

A

OCTAVE PASS ',ANDS 114 CyCLES Pen SECOnt)90 ISO 555 110 1400 1500 5500 ., 11100

N.

\\ Al = 0

,

.. . ..

N \N411,

. /7

"

1''

S1-50

+6

k.

.

FLAT

.

0

,r

,.r N,\

.

,

.N.

,.

,

.

.

-12.C.... _

,

.

_

.

,,

/

f//.

i

63 125.

. ///

250 500 1000a

2000 4000 00011

5 2 2

100 1000FREQUENCY IN CYCLES PER SECOND

t I ry

10000

Figure-5. Allowable octave band sound pressure levels of steady state noises with spectrum slopes of -12; -9,-6°, flat, and +6 dB per octavef or an Al of 0.8 and conversational level speech. The superimposed SI-50contour is a proposed frequency weighting network for evaluating the speech interfering aspects of noise atAl =

'33

20

70 60

1.0

F5 0,8

- 0.6,

0.4

d- 0.2

0,0

70'

RELATION BETWEEN ARTICULATION INDEX (AI) AND

SPEECH INTERFERENCE LEVEL (SIL)

50 40 50 40 50

I I c I

re AI ALA 3 BAND(OSIL) 4 BAND (SIL)12 .5/1/2 -,,5/1/2/4

CAVANAUGH.ETAL / .,. ,,

/

al* fe FRT

60

6 1

40

/ /de 4,' 3 BAND (SIL)

as 01/2/4

f ; 60 70 60 50

PREFERRED FREQUENCY SPEECH INTERFERENCE LEVEL IN DB

40

Figure 6. Relation between Articulation ,Index (Al) and Speech Interference Level (Sil) for 4 noises withspectrum level slopes of -12, -6, flat, and +6 dB/octaife. Three different sets of octaves are shown forcalculating SIL; from left ,to right 500, 1000, and 2000 Hz (.5/1/2); 500, 1000, 2000, and 4000 Hz1.5/1/214); and 1000, 2000, and 4000 Hz (1/2/4). The overall level of each noise is adjusted to obtain Allevels of 0.2, 0.5, and 0.8, and then the SIL is calculated for each of 3 sets of octaves. The data points at anAl of 0.18 are actual experimental points (50% Fairbanki Rhyme Scores) from Klumpp and Webster(1963), i.e.,:these are the Si Ls for noises No. 1, 4, 10, and 15 in their study.

z

Hz--average--or 709 to 2530--overall. or the average measure (SIL-type) the centerfrequencies and bandwidths were 1135 Hz (3.33 octaves) (or an Al of 0.2, 1421. Hz (433octaves) at 0.5 AI, and 1797 Hz (3.33 octaves) for 0.8. These values compare very wellindeed to those proposed in this paper of 1000, averaged over the three octaves 500, 1000,and 2000 Hz; 1428, averaged over the four octaves 500, 1000, 2000, and 0000 Hz; and.2000, averaged over the three o ayes 1000, 2000, and4000 Hz. Cluff also found theSIL -type measure gave §tandard iviations varying from 0.3 to 0.9 (ave.-0.54) while thestandard deviation of the A-wei ed levels varied from 1.6 to 3.8 with an average of 2.20.*

*Cluff, (1969), "A comparison of selected methods of determining speech interferencecalculated by the Articulation index," J. Auditory Res. 9,8148.

34

)

. -The previous analysis has shown th,g,t th,e octaves choasen to _calculate an SIL varyaccording to what Al the SIL is trying to estimate. Webster ('1964a, 1964b) constructed aset, of contours (see Figure 7) for, predicting Als or SILs that also showed the increasing,

importance of the high speed frequencies for increasing levels of intelligibility (and Al). Et issuggested that weighting networks'for sound level meters could be built to predict Al levelsof.0.2 (SI -... 70 'dB); 0.5 (SF= 60 dB); and 0.8 (SI ---: SO dB). A good set of noises on which totest these hypotheses are the 16 noises of Kluriipp and Webster ( 1963).

.

Calculations made on' Klumpp and Webster's 16 noises ,equateth_in,level at Als. of 0.2,0.5, and 0.8, comparing 4 sound level weighting networks A, SI-70, Sl-60, and SI -50, and 3ways of calculating SIL,namely using the 3 octaves 500/1000/2000, the 4 octaves from 500to 4000 and the 3 octaves from 1000 to 4000 are shown in'Table 2. The results generallyconfirm everything that has just been stated, namely, that at a level of intelligibility corre-sponding to ( I) 0.2, the SI-70 and the 500 to 2000 S1L are thebest.(Idwest a and R) (2) 0.5and 0.8the SA-60 and the 500 to 4000 SIL are the best, and (3) 0.8, the SI-50 and the 1000to 4000 SIL Ore good. A-weighting appears slightly inferior- to the proposed S1-60 and anySIL That included 500 Hz. . ,

If the manufacturers of sound level meters -are seriously considering weighting net-works other than A,413,,and C, an SI-60 should be considered: It is appreciably better than Afor predicting speech intelligibility at all AI levels. -

I have shown how the choke of frequencies for SILs or weightin,, networks is:depend-ent on the level of intelligibility to be specified. Now we get back to intelligibility testing.What tests should be used for various levels of Al?

Efficiency factors in test design dictate that the functional relationship between thedependent and independent variable shO4d be steep and lineal' in the critical testing region.Therefore,' consideration should be given to using different language tests for differentcommunication effectiveness areas. For ekarnple,Ifor marginal conditions, AI = 0.2, closedset rhyme words (Fairbanks", 1958; House et al., I 'US; Kreul et al., 196f8; Griffiths, 1967;Clarke, 1965), which yield scores, of about 50%, 'would make ver efficient tests. If alistening sit tionroom or communications equipment required adequate int/elligibility,i.e., an Al o 0.35, then open-set, 1000-word PB tests would yield scores close to 50% andtherefore be- efficient in test design, although inefficient in teRms of crew training, testscoring, etc. The use of closed response -set rhyme words would\be on the:border line pf

... ,'acceptability since the expected scores would be around 75%.

u.At Ai levels around 0.8, no intelligibility test is inherently difficult enough to be anefficient test. Even 1,000 nonsense syllables have an intelligibility" of greater than 90% at AIlevels of 0.8To discriminate between'listening conditionscommunication systems, com-ponents, etc.at Al levels of 0.8_,requires something more than a simple intelligibility test.Reaction times, quality judgments, scores on secondary tests, or interference tasks, such ascompeting messages, have been used or suggested. We will have time to discuss only one ofthese promising approaches, namely the competing message paradigm. Tillman, Carhart, andOls,en (1970) show the decrement in performance or a competing message task due merelyto adding the equivalent of a hearing aid between the sound field and the listenet's ears. Thelistener's task was to recognize in turn one of 50 phonetically balanced (PB) words from aloudspeaker in one Corner of a room while competing sentences at levels 6 or 18 dB dOwnwere coming from a loudspeaker in the other corner 'ahead of the lispner, i.e., the 2

35

2 `)

C

C

.a

-41

110

100

70

BO

80

40

L. di a

OCTINE PAIS SANDS IN CYCKE PIN 3100111241 190 110 ieoa 1000 15oo 11100

I . .e e

. .

13 .

;' S I JO*.

N

SI-60,,,

$ , .a.A

P

SI-50w

.. .

..

....

, , .

91 49, zoo 2000/ 4000 9000.

ticr 100

FREQUENCY IN CYC1000 1-

/IDEft SEC ON

Figure 7: Noise ating contours for estitneting SILs based on different 4eragin ves. e/SI-70 forestimating the /1/2 S1L, = Al of 0.2;/the S1-56 for the .5/1/2/4 S1L, =/A1 of nd th St150 for the

-1/20 SIL, = Al of 0.8. The inverse tifthese)conteurs could be used astfregue y weighti g4ietworks in/ , , r A /

sound level to measure the 7leech\i\nterfering aspcq.ts of, no!:.as. The SI- 0 is the est compromisecontour. .

-0

0G

s

.;

Tab

le 2

,.

-"C

ells

on

the

diag

onal

sho

wtI

ls M

ean

(M),

clai

d D

evia

tion

(a),

and

Ran

ge (

R)

for

the

desi

gnat

ed p

aram

eter

s fo

r A

l lev

els

of(f

rom

left

to r

ight

) 0.

2, 0

.5, a

nd 0

.8. C

ells

ove

die

-dia

gona

l\sho

w th

eph

ysic

al s

tatis

tics

for

the

16 K

lum

ppan

d W

ebst

er (

1963

) no

ises

(AI

is n

ot a

fac

tor)

. The

num

bers

in th

eel

ls (

abov

e th

e di

agon

al)

repr

esen

t the

M, 0

, and

R o

f th

e di

ffer

ence

dis

trib

utio

n be

twee

n th

epa

ram

eter

s lis

ted

at th

e le

ft a

nd a

cros

se

top.

',..

....

-.3.

-,--

---_

__-

-,W

eigh

ting

etw

orks

SIL

s.

,-

A-W

tng

A-W

tng.

SI-7

0_

S1-5

03L

43H

..5/1

/2.5

/1/2

/41/

2/4

83.1

72.3

62.6

3.7

2.8

0332

11.3

-8.

39.

7 ,

N

5.6

2.6

10.0

"

77.4

66.6

56.9

2.4

3.1

4.0

9.0

9.0

11.5

6.6

2.1

10.0 1.0

2.8

9.5

-3.

24.

414

.7 2.4 8

18.9

N

76.5

65.7

55.8

3.0

1.6

2.3

11".

07.

610

.6

3.3 3.2'

9.7

r2

9:8

3.3

9.3

4.2

1,.9

81

3.2

1.9

6.8

10.6

3.2

11.5

k5.

0,3.

512

.7

11.6 4.8

16.4

6.0

Ms.

--5.

819

.7R

4.1

1.5

6.2

79.7

68.9

59.2

5.6

3.1

2,6

16.5

9.8

7.8

14.6

7.4

2.7

8.2

73.0

62.

352

31.

51:

62.

7'63

6.4

10.4

0:8

"-

1.7

6.9

72.2

61.9

51.6

3.0

0.7-

1.40

.,9.

44.

56.

6

5.0

-3.

2a

11.5

8.31

r

0.9

'a

5.2

13.0 1.0

2.2

8.1

71.2

60.4

5.4,

3.2

17.4

10.9

5,0, R

50.6

.M

2.7

a9.

4R

-

to

ti

D

loudspeakers were 45°,,to the left and to the right 15r the listener's nose. Unaided listeningoyass( I) binaural; F2) monaural direct, in which the speedh was on the side of the liliening

,,,,

ear, and the others ear was occluded with a muff; and monaural indirect, in which the speechwas on the side of the occluded ear. Aided listening used an artificial head in the sound fieldWith two hearing aids and connections via arhplifiers and calibrated attenuafors to insertearphones in the ears of the remote listener:Again three conditions were testedbinaural,

. ,monaural direct, and monaural indirect.

Four groups of 12 subjects each were tested including (1) those classified audio-logically as normal (average age 22); (2) With moderate hearing losses diagnosed as conduc-ive age 42); (3) sensorineural (averdge age 51); and (4) presbyacusic (average age

70).. The grotips will be indicated by N, C,1/45 and P, respectively.All listening Avas of a level ,30 dB above the threshold for spondee words, 30 dB

,Sensation Level (SL), under each,Of the 6 conditions. Figure 8 shows the results which canbe summarized as follows: Cor4ared to an earlier reference group of 20 no al hearingsubjects, on the PB word/sentrice competition task (Northwestern Univers'ty Auditory

c, . .

lest 2, Carhart et at., 1963, ,the .N and C groups sitting in the s and field (i aided) heardessentially at reference le ef; the; S and P groups required, o .average, 14 dB better

ord-to-sentence differe lal than the N and-C groups in th sound fie d; the N group/required about the sam increase in word-to-tehtence differ& tial when hearing aid was' interposed between them and the sound field; the C group required an e en greater increase\ .

. i. for the aided conditionsiabout 18 dB more; and the S and P.groups, who required a 14 dB

word-to-sentence (W/S) improvement in the .unaided case, required further improvementswhich increased as the basic word-Co-sentence (W/S) differential increased. Rettated, the S

. and P groups are worse off than the N and C groups in listening to competing speech signals30 dB above their speech threShold, whether' listening with or withoukhearing aids.

These results show both a hearing deficiency penalty and an equipment-imposedpenalty when listeners are placed in competing .message,listening conditions. This is badnewsor people incurring noise-induced hearing losses whO are generally sensorineural innature. No only do they have more difficulty than theif normal-hearing or conuctivelys.deafened f 'ends in cocktail party environments, but t ey cannot look forward to a heariaid to he equalize theii. relatiye disadvantage. / / tv

7 last/point I W ant to; make concerns lis ening--Ato Speech in noise/ bile wearingear p T or muffs It as I2ng b n estayifshed t t ilynoise levels greater an 90 d13speecis ayd/better when/Wearing aring prote-eti n. Tkis early work of ryter (1946) was,young no al-hearing subjects.. Howev , there, at least ,one study' by FrohIich 970)

twhich shoWs that/unlike young no 1 hearing males, senior aviators with high-fr uency ,isensorineUral loss ado not disc . mate digits better in noise-levels above 100 13 when

i,w wring good noise-dttenuating/ar muffs., He shows that this could be expected by plotting/;cTh afing-level and hearing-lev3lAtinde,r-muff for senioraviators on the spte h are4,,and no' emasking area.' This procedde shows that the! muff cuts out a regi2r( speeC 4req nctes)

...,

/ where the'speech is well above the masking 'rise: It seems saf ,to y that., -traumai A

lisle ersZ have more difficulty than normals in discriniinatin speeckt in/quietenjoyed /

an

p icularly in competing message situations. They do not get the full ben fit enjoyed/ ynormal listeners of increased intelligibility, i high no s by wearing hearing protect ;sand

they cannot expect a hearing aid to help them unt gle competing messages;

\,

A

38"

a

fg 4000 90o 801--.1

*-4L.L.1

60.

50.

at, 4011.0

39i/

20i--, ,

-\-...p . LU 1 0V/-1-1.14 w

-a..

d

9

-12 -6

EFECtIVE WO ENT/E

Rn.7

Figure 8. Percy ge of 50Pts words correct itt,the presence o competing sentishces at various word-to-Sehte ce 'd'. erentials. The paramq are REF=20 referen normal hearing ,listener-NIC normal or ,/ conductillt4 pathology experimental listeners; P/8 presbyacustc or nsorineural pathology experimental s'/ listen r..Aided refers to listening via hearing -aid cir9li - Unaided rlferf/o listening normally in a sound '4

4.

7 , -4/treated roontoFrepth Tillman, Carhart, and ()keg (197

,A1, / r /-,),An SU ary, I ,have tried t /telt you in(tis presentation that the/ octaves chOien 'to?.

.7

4 alCUlateVie SI and /or the w hting networks that could be builtto a soundlevel rr4teto measure the i terferece noise with speech vary a' a func of whatleyel of,speechcommunication ou desir to design forfora Correspondi gly, the tests'youru,sto evaluate aListener" or la sys ern_ At in the' same manner, senteric intelligibility tes s being best for a,...

3 i.:, .

I l i

tii \

O

basically bad system, Word or nonsense syllable tests for a good system, and competingmessage tests or judgment tests for an excellent system. Persons with noise-induced hearingloss cannot hear as well as normals when wearing plugs or muffs in moderate to high levels

"" of noise nor can they by wearing a hearing aid unscramble competing messages (at a cocktailparty) as well as normals,

References«

I. - House, A.S., C.E. Williams, M.H.L. Hgcker, and K.D. Kryter (1965), "Articulationtesting methOds: Con.sOnantal differentiatio4ith a closed-response set," J. Acous.Soc. Amer. 37, 158466,, also USAF EST-TDR 63-403.

2. Clarke, F:R. (1965), "Technique for evaluation of speech systems:: Final Report ofStanford Research 'Institute Project 5090 on U. S. Army Electronics Laboratory Con-.tract DA 28-043 AMC-00227(E), August 1965.

3. Kreul, E.J. Nixon, J.C., Kryter, K.D. Bell, D.W. Lang, J.S. and Shubert, ED. (1968),"A Proposed clinical Test of Speech Discrimination," J. Speech and Hearing Res. I I,536-552.

4. Fairbanks, G. (1958), "Test of phonemic differentiation: The rhyme test," J. Acqust.Soc. 'Amer.,30, 596-600.

*41

Egan, Jy.,(1948), "Articulation testing methods," Laryngoscope 58, 995,;991.

. Wekster, J.C. (072), "Compendium of speech testing material and typical noisespectra to be used in eValftating communication equipment,".NELC Tech'Doc. 191 Of13 Sept.

7. Clarke, F.R., J.C. Nixon, and S.E. Stuntz, (1965), "Technique for evaluation of speechsystems," Stanford Research Institute Semi-annual Report of SRI Project 5090 fqrU. S. Army Electi-onics LaboratorySontract DA 28-043 AMC-00227(E) AD 462836.

8. Dreher, J.J. an J.J. (1957), "Effects of ambient noise on speaker intelligibility,for words and p rases:: J. Acoust. Soc. Amer., 29, 1320-1327.

Pickett, J.M. ( °>'956), "Effects of vocakforce on the intelligibility of speech sounds," J.Acoust. Soc. Amer., 28, 902-905.

9.

10. French, N.E., and IC. Steinberg (1947), "Factors governing the intelligibility ofspeech sounds," J. Acoust. Soc. Amer. 19, 90-119.

Beranek, 'L.L. (1947a), "The design of speech communication systems." Proc. Ihst.Radio Engrsr, 35,880-890..

9

40

27

4

7'I12. Krytur7,K.D., Methods for thetalculatio.n and use of the articulation index. Jy, Acoust.

Soc. Amer., 34, 1689-1697 (1962a).

13. Kry ter, K.D., Validation of the articulation index. J. Acoust.1698-1702 ( I 962b).

Soo. Amer., 34

14. Webster, J.C. (1969), "Effects of_noise on speech intelligibility,"Conference on Noise as a Public Hazard. Proceedings, 13-14 JunSpeech and Hearing Association (ASHA) Reports 4. ;

49-73 in National68, TheAmerican

15. Flanakan, J., and H. Levitt (1969), "Speech "interference froin community noise", p-7 167-174 in National Conference on Noise as a Public Health Hazard, Prodeedings,

13-14 dune 1968, The American Speech and Hearing Associatign, ASHA reports 4.

16. Beranek, L.L. (1947b), "Airplane quieting 11.- Specification of acceptable noise levels."Trans Amer. Soc. Mech. Engri, 69, 97-100.

17. Botsford, J.1-1. 1969), "Using sound lePels to gauge human response to noise," Snd.and Vib. 3(10) 6-28. vow%

ii

18. Webster, J.C. ( 965), "Speech communications as limited by ambient noise,'.' J.Acoust. Soc. A er., 37, 692-699.

19. Cavanaugh, W.J., W.R. Farrell2P.W. H-irtle, and B.G. Watters (1962), "Speech Priva6/'in Buildings", J. Acoust. Soc, Amer. 34, 475-491

20. Webster, J.C. (1946a), "Generalizes speech interference contours," J. Speech andHearing Research, 7, 135-140.

21. Webster, J.C, (1964b). "Relations between speech-interference contours and idealizedarticulStion'index contouts,".J. Acoust. Soc. Amer., 36, 1662-1669. 4.

'22., Klumpp, R.G. and .P.C./Webste01963), 'Physical Measurements of equally speechinterfering Navy noise,'kl. Acous. Soc. Amer. 35,1328-1538.

23. Griffiths, LI). (1967), "Rhyming minimal contrasts: A simplified diagnostic articu-lation test", J. Acoust. Soc Amer.-42, 236-241.

24. Tillman, T.W.,/11. Cathart and W.O. Olson (1970), "Hearing al efficiency in a com-peting speech situation", J. Speechand Hearing Research 13, 789- 11..

25 Carhart, R,., T.W. Tillman and- L. Wilber (1963), "A test for speech discriMination -composed of CNC monosyllabic words," Percsplual-and Moto Skills 16:680.

41

26. Kryter, K.D. (1946), "Effects of ear protective devices on the intelligibility of speechin noise," J. Acoust. Soc. Amer,. 18, 413-417.

27. Frohlich, B. (1970), "The effects of ear defenders on speech perception in militarytransport aircraft," North Atlantic Treaty Organization (NATO) Advisory)Group forAerospace esearch and Development (AGARD) Advisory Report 19.

Webster, J.C1 and C.R. Allen (1972), "Speech intelligibility in,,,,Naval Aircraft radios,".NELC TR 1830 of 2 Aug.

4

42

25