Acoustic survey of an open-plan landscaped office

Publisher’s version / Version de l'éditeur:

Vous avez des questions? Nous pouvons vous aider. Pour communiquer directement avec un auteur, consultez la première page de la revue dans laquelle son article a été publié afin de trouver ses coordonnées. Si vous n’arrivez pas à les repérer, communiquez avec nous à PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca.

Questions? Contact the NRC Publications Archive team at

PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca. If you wish to email the authors directly, please see the first page of the publication for their contact information.

https://publications-cnrc.canada.ca/fra/droits

L’accès à ce site Web et l’utilisation de son contenu sont assujettis aux conditions présentées dans le site LISEZ CES CONDITIONS ATTENTIVEMENT AVANT D’UTILISER CE SITE WEB.

Internal Report (National Research Council of Canada. Division of Building

Research), 1972-11-01

READ THESE TERMS AND CONDITIONS CAREFULLY BEFORE USING THIS WEBSITE.

https://nrc-publications.canada.ca/eng/copyright

NRC Publications Archive Record / Notice des Archives des publications du CNRC :

https://nrc-publications.canada.ca/eng/view/object/?id=8f7241a6-3d11-4bd5-bd50-43924b4a64c9 https://publications-cnrc.canada.ca/fra/voir/objet/?id=8f7241a6-3d11-4bd5-bd50-43924b4a64c9

NRC Publications Archive

Archives des publications du CNRC

For the publisher’s version, please access the DOI link below./ Pour consulter la version de l’éditeur, utilisez le lien DOI ci-dessous.

https://doi.org/10.4224/20359099

Access and use of this website and the material on it are subject to the Terms and Conditions set forth at

Acoustic survey of an open-plan landscaped office

DIVISION OF BUILDING RESEARCH

ACOUSTIC SURVEY OF AN OPEN-PLAN LANDSCAPED OFFICE

by

A. C. C. Warnock, D. N. Henning, and T. D. Northwood

Internal Report No. 400 of the

Division of Building Research

OTTAWA November 1972

)

ACOUSTIC SURVEY OF AN OPEN -PLAN LANDSCAPED OFFICE

by

A.C.C. Warnock, D.N. Henning, and T.D. Northwood

PREFACE

The Department of Public Works, Capital Region, cornrni e s ioried an acoustical study of the new landscaped office ac cornrnodat'ion of the

Ministry of Transport, Tower "C", Place de Ville, Ottawa. The office staff taking part in the expe r irne nt included both professional and clerical

workers. They had been previously ac comrn od ated in older buildings with cellular type offices each containing as rn arry as four workers.

The study was carried out in three parts each part analyzing an area of intere st.

a) A subjective a s se s sment of the ITlasking sound installation by the occupants was obtained by exposing therri to two different sound spectra each day and polling t hern for their opinions. b] A direct rne a su r ern.errt of the effectiveness of a rn a sking

sound sy stern in iITlproving speech privacy conditions was rriade by rn e an s of subjective articulation tests. These subjective rn e a su r ernent s were cornpa r e d with objective e stirna.te s derived f'r om physical rne a su r erne nt s ,

c) Objective rn e a au r errre nt s of sound propagation in the land-scaped office environment were rn ade and the efficacy of several e Iernent s used to decrease sound propagation was assessed.

LANDSCAPED OFFICE

by

A. C. C. Warnock, D. N. Henning, and T. D. Northwood

In the last few years the open-plan office has be corrie increasingly popular (1 to 3). There are undoubtedly significant advantages in this use of open space, but such offices often have acoustical pr ob lern s and psychological studies have seldom. investigated the acceptability of the acoustical environm.ent. The survey now reported was undertaken to exam.ine these p r oblern s in an actual situation by evaluating the acoustics of the space.

A landscaped office usually include s a large work area (recom.-m.ended absolute rniriirnurn 4000 sq ft) fitted with a carpet, an

acoustic ceiling, air conditioning, uniform. glare -free lighting and m.odern functional furniture. Sound-absorbing, free-standing screens, plants, objets d'art, as well as the use of colour, texture and various planning approaches provide the necessary visual privacy and add variety and interest for the occupants. Figure 1 gives the general plan of part of the space under study. Induction units installed around three sides of the office for heating and cooling, com.plicated the

acoustic tests to som.e degree, as will be explained.

In such an open space conversation and other noises rnay propa-gate freely over large areas and still rem.ain intelligible and distracting. The provision of sound-absorbing flooring, ceiling and furniture is

therefore a reasonable m e an s of reducing sound propagation and partially alleviating the pr oblern, Such me a sur e s , however, do not provide full acoustical privacy. Confidential conversations m.ay still be overheard at adjacent work positions. Careful arrangem.ent of

furniture, spacing of occupants and use of acoustical screens can help(4). but it has be c orne alrn o st standard procedure to install an electronic m.asking noise sy s'tern , An array of loudspeakers is placed out of sight in the plenum. above the ceiling. A noise generator and filters provide noise with a spectrum. shape chosen, to a large extent, by the supplier of the sy stern , The noise signal is fed into the

loud-speakers and a relatively uniform sound field established in the plenum and in the room below. It is the purpose of this system to raise the background noise level by an amount sufficient to mask the intelligi-bility of speech from nearby work positions without causing undue dissatisfaction am ong the office per sonnel because of the increased sound level.

­2-An  array of  ceiling  baffles  m.ay  also be  used  to  reduce  sound 

t r an srn.i s s i on ,  These  are  intended  to  increase the  effective  absorption  of  the  acoustical  ceiling,  and,  by breaking  up  the  plane  surface,  to  interfere  with  specular  reflections  from.  the  ceiling  to  neighbouring  workplaces. 

The  pre sent  study has  exam.ined  the  acoustic  environm.ent  of  a  typical  landscaped  office  to  determ.ine  the  following: 

subjective  preferences  in  level  and  spectrum.  shape  of  m.asking  noise; 

effectiveness  of  various  m.asking  noise  conditions  in  m.asking  speech; 

effect  of  proposed  ceiling  baffle  configuration  on  sound  transm.ission  and  speech privacy; 

sound  propagation  characteristics  of  the  space;  effectiveness  of  the  acoustic  screens. 

The  experim.ents  perform.ed  are  described  in the  following   sections.  

SUBJECTIVE  REACTION  TO MASKING  NOISE 

As  introduction,  the  potential  benefits  of  a  m.asking  noise  system.  were  explained to  the  participants  and  their  as sistance  was  sought  in determ.ining  optim.um.  working  conditions  (Appendix A).  The  subjects  were  therefore  aware  that  m.asking  techniques  were  in  use  and,  although  exact  details  were  withheld,  they were  under  the  im.pression  that the  m.asking  system.  was  functioning  at  all t irne s , 

Four  different  spectra were  used  at  three  pos sible  sound   levels In addition to the  existing  noise  from.  the  air  conditioning  

sy stern ,  Spectra  are  de signated  as  follows:   1.  3dBj octave  reduction  with frequency 

2. 5dBJ octave  II II II

3.  7dB/ octave  "  "  " 

4.  sim.ilar  to  No.  3  but  reduced  high  and  low  frequencie s ,  These de scriptions are only  appr oxirnate  and  the  m.easured  spectral distributions  are  shown  in  Figure  2  (a  to  d).  For  com.parison  a  spectrum.  shape  recently  recom.m.ended  by  Beranek and  others  (5)  is  shown  in  Figure  Ze , 

Sound levels are designated as follows: A-5ldB{A); B-48dB(A); C -45dB(A). Thus, B2 denotes -5dBj octave and a weighted over -all level of 48dB(A). The test schedule is shown in Table I.

For a period of two and one half weeks the occupants were exposed daily to two background noise conditions, one in the

morning and one in the afternoon. The change -over took place at lunch time when the space was vacated. Each day a questionnaire (Figure 3) was completed and collected in the late afternoon. The subjects stated their preference for either the morning or afternoon condition and provision was made for additional comments. This series of tests was followed by a two-week study in which each con-dition remained constant for one week. A one-day test was then conducted with the air -conditioning diffuser s turned off; a low-level uniform sound field was then created. The opinions of the occupants on these last two tests were obtained during an interview held one day later. Finally, when the formal study had been completed, one further unannounced experiment was made.

Details of the subject group and their work functions are set out in Table II. Of the 40 people who worked in the office some were on outside duty most of the time, so that about 30 people answered the questionnaire on a regular basis. The number of cards returned each day varied from 22 to 35, with an average of 28 cards.

Occupantsl _Comments

Generally, the occupants disliked the background noise system and preferred the ambient condition. Of the 124 comments received, 81 per cent indicated negative reaction to the noise or relief when the

system was turned off. There was almost no statement that could be interpreted as favourable to the noise system.

Some of the comments follow:

I} "Noise is still too loud for comfort of co ness to efficient work" (5ldB(A)}.

ncentration and

conducive-2) "This afternoon is not tiring to hear" (N afternoon).

o artifical noise in the

3) "Turn it off" (5ldB(A)).

4) "Noise bothers melt {45dB(A}}.

5) "When artificial sound system was turne total relaxation felt throughout my body and annoying all morning" (48dB(A)}.

d --off at 12: 30 a system was t sense of oo loud

-4-6) "The noise continues - - but to a lesser degree than some days. The noise level still remains too high and is unacceptable to me, and it does not contribute to ideal working conditions. Read the October 1971 issue of the Readers Digest -- page 120 "I am J oer s Ear" and do not subject us to any more noise please" (45dB(A) and 48dB(A)).

In many cases a choice for uno preference" was qualified by a sugge stion that both conditions were unde sir able. The choise of "slightly prefer morning" probably means "this afternoon seems especially bad so the morning must have been a bit better". The choice of "slightly prefer afternoon" probably means "it does not seem quite so bad as usual this afte r noon ";

SOUND LEVEL PREFERENCE

There was a consistent preference for lower sound levels. In

considering all the half-day comparisons of sound levels, 59 per cent preferred the lower level, 30 per cent gave no preference, and 11 per cent preferred the higher level. As the difference between sound levels increased, so did the strength of preference. Eighty-five per cent of the returns favoured ambient over 5ldB(A); 48 per cent favoured ambient over 48dB(A). The detailed results are shown in Figures 4 to 8. Figure 9 shows the mean for all sound level

comparisons.

Many occupants were unable to remember much about the weeks of 1 November and 8 November. The results in Figure lOa exclude 13 who could not remember and one who preferred the quieter week but could not remember which it was. If the 14 excluded are included as no-preference votes, the results change to those shown in Figure lOb. SOUND SPECTRUM PREFERENCE

Although the results show only slight preferences in some cases, the wide range of responses to similar conditions makes any con-clusion very tentative. At 48dB(A), 5dBJ octave is preferred to

7dBJ octave. At 45dB(A} and 5ldB(A), 7dB/ octave is preferred to 5dBJ octave. There was no preference between 3dBJ octave and

5dB/ octave at 45dB(A). The detailed results are shown in Figures 11 to 14.

PHYSICAL MEASUREMENTS OF MASKING SPECTRA

Masking spectra 1, 2, and 4 were generated using a Bruel

& Kj ae r noise generator and multifilter. Spectrum 3 was that set up by the supplier of the sound masking system. The spectra were

-5-adjusted and measured at the approximate centre of the te st space and at a height of about 4 ft (see Figure I for test area layout). The signal from a l v in , B & K condenser microphone was passed through a microphone amplifier into a General Radio real-time analyser. The spectra so obtained are shown in Figure s 2 (a) to (e). One unfortunate feature of the study space was the pre sence of fairly noisy induction heating units around the three perimeter window walls. The noise from them created a gradient in the ambient background level. the sound level being highe st. about 4 7dB(A). at the windows and dropping toward the core wall by as much as 7dB(A). Although this disturbing effect was present. however. it did not seriously alter the shape of the masking noise spectra except for a few Cases where the high frequency components were slightly increased.

PROXIM.ITY TO DIFFUSERS

In investigating the possible influence of diffuser noise the subjects were considered in two groups:

those located at the building perimeter; and those located in the central core area.

This division not only separted the subjects close to the noise

generated by the air-conditioning diffusers from those further away but also separated senior personnel from lower ranks. In all but two Cases. one of which ocurred on the first day. the response from the perimeter group Was similar to that from the group in the central part of the building. People located at the perimeter were not so strong in their opinions, but because the number of returns from the perimeter group Was often quite low this conclusion is tentative.

Unannounced Experiment

When people are aware that a rr.. asking system is pre sent in an open office there is a tendency for some individuals to attribute various ailments to the effects of the noise. for example. headaches, nervous tension. inability to sleep. ringing ears. As the levels in use are not in the least harmful. it is apparent that the noise is a target for emotional and psychological discontent. It is generally

advised. therefore, that the presence of a masking sound system not be announced. In inve stigating this aspect of open office planning, a further clande stine experiment Was carried out. At the completion of the initial phase of the experiment the staff were informed that in view of their rejection of masking sound they would not be subjected

-6-to it any longer. Four rn orrth s later the rna sking noise was turned on at night at about 1

1/z

dB(A) above the arnb ie nt level in the middle of the office. At intervals of a pp roxdrnat e Iy one week the

sound was increased in steps of 1 l/Z dB{A). A rn ernbe r of the office staff was aware of this experiInent and rn onito r ed office conversations for s orne evidence of awareness of the noise. There was no staff reaction until the level at the rnid -point of the office reached 47dB(A), when sorne c orrrpla'int s were rnade ,

During the period when the sound was off, no cornpla int s about lack of privacy were rnade , This, together with the second

rejection of m a sking noise, very strongly suggests that this partic-ular group does not need privacy and that rna sking noise is a rn o st unwe lcorne intrusion.

CONCLUSIONS

This section of the survey m ight be described as a partial failure ina smu ch as no preferred spe ct rurn choise was det.e rrrrined , It was not without value, however, because several irnport ant points emerged:

The original high levels [5ldB{A) ] set by the supplier prior to occupation of the space and used in the early parts of the test provoked surprisingly hostile reactions and this unpleasant pre l irrrina.ry experience rn ay have disposed the subjects to view the whole idea in a negative

way. The result of the unannounced expe r iment , however, tends to indicate that the rejection is genuine. Although the comrne nt s and reactions became less ext r erne as time went on, they r ernarned negative. The reasons for this consistently negative attitude be c am.e slightly clearer during the interviews at the end of the

survey. From their comments, the people in the area apparently felt no great need for acoustic privacy and did not, therefore, require any masking.

The e xpe r irne nt a l result for 18 Novernbe r when the induction units were turned off seems to be a definite rejection of noise, although it was only set at 45dB(A) which is less than the induction unit noise in rn o st cases and theoretically less obtrusive in character.

It should be erripha s i sed that a cornpl.etely different result can be obtained where the need for privacy exists. A srnaIl scale e.xpe r irne nt (6) in another open office situation where the staff had a high speech privacy r equi r erne nt resulted in unequivocal acceptance of a m a sking sy stern operated at 48dB(A) with a spe ct rum shape similar to Spe ct rurn Z. It is therefore imperative that privacy requirements be evaluated at an early stage of planning so that the need for sound masking can be estimated.

If privacy requirements are sufficiently stringent, it rnay be advisable to reject open planning and to use a cellular type of

st r u ction ,  As  this  experiment  shows,  however,  the  need  for  speech  privacy is  often  not  so  great  as  was  initially believed. 

The  two­week  experiment  seems  to  indicate  that  48dB(A)  is  not  too  intrusive;  many  people  were  unaware  of  it.  (The  un-announced  experiment  points  to  a  lower  acceptable  value  of  about  46dB(A),  which  could  be  used to  irnpr ove  speech  privacy without  irritating  staff. 

Experiments  elsewhere  (7)  have  shown that  the  maximum  advisable  masking  level  is .about  48dB(A).  When  asked to  speak  at  a  normal  level against  varying  background  noise  levels,  people  begin  to  raise their  voices  when  the  background  noise  exceeds  NC  40,  which  corresponds  to  about  48dB{A).  Apparently,  it  is at  this  level  that  the  noise  begins  to  be  intrusive.  Further  increase s  in masking  level  are  not  so  effective  as  might  be  expected because  of the  rise  in  voice  level  and  because  higher  levels  tend  to  cause  some  dis satisfaction •. 

It  is  concluded that  if  a  masking  system  is  deemed  necessary  it  should  not  be  operated above  48dB(A)  and  that  staff  should  not  be  informed  of  its  existence. 

MEASUREMENT  OF  INTELLIGIBILITY  OF  SPEECH 

To  estimate the  effectiveness  of  different masking  spectra  it  is  possible  to  make  calculations  based  on  assumptions  about  voice  spectra  and  ambient  noise  levels  (8).  These  procedures  were 

developed,  however,  for  a  different  purpose:  the  evaluation  of  speech  communication  systems.  It  was  considered  desirable  in  the  present  application to  compare  the  results  of  such  calculations  with  direct  subjective  articulation measurements,  utilizing  a  test  crew listening  to  and  trying  to  identify  a  list  of  selected words.  The  test  chosen  was  that  described by Griffiths  (9).  It  requires  no  subject training,  has  no  learning  effects,  and  the  scoring  is  quite  straightforward.  Calculation  of Articulation  Index 

Thirteen te st  positions  were  chosen  (Figure  1)  and  the  articulation  index in  each was  calculated  as  follows: 

1)   The  noise  level  in  each te st  position was  measured  in  dB(A)  for  each masking  condition. 

2)   The  ideal  speech  spectrum  (8)  in Figure  15  was  used  to  calculate  articulation  index  (AI)  as  a  function  of  signal­to-noise  ratio (SiN) for  all  the  spectra used  (Figure s  16  a,  b).  The  long  term  rms  value  of  ideal  speech was  taken  as  65dB. 

ｾｾ   ＭｾＭＮ｟ｾＭｾｾｾｾｾＭＭＭＭＭＧＭＭｾｾｾｾｾｾｾＭＭＭＭＭＭＭＭＭＭＭＭＭＭＭＭＮＬ［ＮＬＮ  

­8-3}   Using  the  values  of  signal  attenuation  obtained  in the   second  series  of  propagation measurements  (Section IV),   the  signal  level  in  each test  position was  determined.   4)   The  information  obtained  in  steps  (1)  and  (3)  gives  a  value  

of 

siN

for  each test position,  and  Figures  l6(a)  and  (b)   derived  in  step  (2)  can then be  used  to  give  a  preliminary   value  of  AI.  

5)   These  preliminary values  of  AI were  corrected to  take   account  of  azimuthal  effects,  using  Figure  17  (9).  The   number finally  obtained  for  each  position  is  called  the   calculated AI  (Table  III).  

SUBJECTIVE  INTELLIGIBILITY  ASSESSMENT 

In the  G'r iffi.th s!  test the  listeners  are  presented with  a  test  list  containing  250  CVC  (Consonant­ Vowel­Consonant)  words 

arranged  in  rhyming  groups  of five.  The  words  within  each group  of five  differ  in either  the  initial  or  final  consonant,  but  not  in  both.  The  speaker  reads  one  word from  each group  of five  and  the  listener  has  to  make  a  choice  from the  five,  guessing if he  does  not  know.  This  forced  choice  from five  gives  a  chance 

score  of  20  per  cent  on the  test. 

A  volunteer  crew  of  twelve,  drawn from  outside  the  occupants  of the  space,  comprised the  test team.  Each was  given  an  audio-metric test  and  all  had  hearing  that was  within  normal  limits 

considering  age  effects.  The  tests  were  carried  out  in the work space,  during  the  lunch hour.  As  the  team was  not  specially  selected  there  was  a  large  scatter  in the  results,  reflecting  different  aural acuities  and  listening  skills. 

The  mean  subject  score  for  each  of  the  positions  shown  in  Figure  1  was  obtained  under  the  different  noise  conditions  in the  schedule.  The  voice  level  of  the  speaker  was  monitored  at  a 

distance  of  3  ft  using  a  B  & K  sound  level meter  on the  A  scale  and  the  fast  time  constant.  The  initial  tests  were  made  with the  observed  peak  reading  on the  meter  at ＶＰ､ｂｻａｽｾ  but  it was  decided  that 

this  was  too  high  and  during  the  latter tests  a  peak  reading  of  55dB(A)  was  maintained.  Due  to  illness,  the  speaker  unfortunately  had  to  be  changed  towards  the  end  of  the  test  series,  but  repeat  tests  indicated  that the  effects  of  this  change  were  of  minor  importance.  Table  IV 

shows  the  mean  score  obtained  in  each position for  the  60dB(A)  tests  and  Table  V  shows  those  for  the  55dB{A}  tests.  The  relative  effect-iveness  of  the  different  masking  spectra can be  estimated from  these. 

Comparison may be made more easily, however, by the mean normalized score s shown in Figure l8{a) and (b). The te st score in each position was normalized to the score obtained under

ambient conditions with a 60dB(A) voice level. The normalized scores for all positions were then averaged to obtain the mean normalized score.

The expected mean normalized score was obtained from the calculated AI value s for a typical po sition (T able III) and Figure s 20 and 21. To take account of the different voice levels used in the tests, the AI values were increased by 0.17 when a 60dB{A) peak voice level was used. The score s so obtained were

normalized as before t o give Figure l8{b). It is quite clear that by choosing a poor spectrum workers in the office can be

subjected to fairly high noise levels and gain little extra privacy. Conversely, a wise choice of masking spectrum Can greatly improve speech privacy without seriously degrading the environment.

Figure 19 shows a plot of the mean score for each te st in each position versus calculated AI; the different voice levels were allowed for as before. The two lines are simply linear regression lines and there is no a priori reason to expect them to fit. The

standard error, Sy' is 10.57.

[S2

=

ｾ  

(y observed - y calc) 2/ n

J

y I

and the correlation coefficient between mean score s and calculated AI is 0.82.

Two othe r curve s from diffe rent source s are of inte re st in relation to this plot. The first is obtained from Griffiths' work (8) by assuming that when AI

=

0, the score obtained is 20 per cent, the chance score. This necessitates a small amount of extrapolation of Griffiths'data. The curve can best be fitted to the points of Figure 19 by increasing all the calculated AI values by 0.04. When this is done the standard error, Sy has a minimum value of 10. 33 and the index of correlation p of 0.83

(o 2

=

1 _ S

ｾ  

cr

ｾ  

) •

The result of this fitting process is shown in Figure 20.

The second curve of interest, Figure 21, is obtained from the 1000 PB (phonetically balanced) word curve in the ANSI standard (8) by as suming that the subject hears p per cent of the words correctly and guesses one fifth of the remaining (100 - p) per cent correctly. (Although this treatment is rather unsophisticated, it appears to give an approxi-mately correct cu r ve , ]

ＭＭＭｾＭＭＭＭＭＭＭＭＭＭｾＭＭＭＭＭＭＭＭＭＭＭＭＭＭＭＭＭＭＭＭＭＭＭＭＭＭＭＭＭｾＭＭＧＭＧＭ

-10-To obtain a be st fit of this curve to the points of Figure 19 all the calculated AI values are decreased by 0.05 so that the m.inirnum. value of Sy

=

10.85 and p :II 0.8l.

It is necessary at this point to elaborate on a discrepancy in the foregoing work: the ideal voice level assum.ed for the calculations of AI was 65dB long -term. rm.s whereas the real voice level was

closer to 55dB long-term. r-m s , If taken into account this difference of 10dB necessitates a reduction in the calculated AI values of 0.33. The change, if rriade , would m.ean that when AI is zero, subjects can score 50 per cent on the Griffiths' test. This result seem.ed som.ewhat surprising and because of this, and other reasons, it was decided to rnake a careful com.parison of the Griffiths! test with the standard ASA te st.

This com.parison has since been m.ade in the DBR/NRC labor-atory using a carefully selected test crew with good hearing. The

scatter in the results is m.uch less but the sam.e feature erne rg ad; when the calculated AI was zero, the score obtained on the Griffiths! test was 60 per cent.

To achieve acceptable privacy in an open office, it appear s that AI should be less than O. 15. The critical range of AI is from. 0.35 down to zero. In this range very srnall change s in AI produce great change s in sentence intelligibility, and any te sts used to evaluate AI in this region should be either very sensitive to sm.all change s in AI or

very accurate. The Griffiths' test is not at all sensitive to AI changes and, when used with a random.ly selected crew, is not at all precise. The precision of the te st can be im.proved by selecting the te st crew with care, but this would erase the sim.plicity of the test. One m.ust conclude that the Griffith s! test, although superficially attractive, is not suitable for simple estim.ates of Al in the open-plan office.

Re suIts

The findings of the intelligibility tests together with AI calcula-tions provided a m.eans of predicting the speech privacy relation corre s-ponding to any talker-listener pair in the space. As an exam.ple, a

range of conditions for a talker-listener separation of 12 ft is sum.m.arized in Table VI. (Som.e of these values have been adjusted to m.ake them.

directly cornpa r ab le , ] It will be seen that with no added m.asking noise (natural am.bient 40dB{A)) the Articulation Index varie s from. 0.35 to 0.88 depending on the arrangem.ent of screens and talker orientation. In the be st case speech privacy would be rated as "very poor". With the 50dB{A) of m.asking noise the be st case is prom.oted to "excellent". The fact that so m.any of the ratings are not satisfactory is a result of the poor perform.ance of the ceiling installed.

It should be remarked that the se privacy criteria relate to how much a listener may understand if he is actively paying atten-tion. The converse question of how much he will be disturbed by intruding speech is more complex, but probably the acceptable level will be somewhat higher. It is surmised that the listener who is typically exposed to several more or less concurrent voices in the AI range of 0.30 to 0.50 will be relatively undisturbed by any one of them.

PHYSICAL PROPERTIES OF THE SPACE

The general behaviour of sound in the test space was examined by means of propagation measurements taken with the apparatus

indicated in Figure 22. The test signal was pink noise (wide-band random noise containing equal sound power per third octave). The sound level meter was set on either a flat frequency response or the A-weighting network.

The first series of te sts involved a propagation path with no intervening screens except those introduced to investigate screening effects. The se measurements, to be discus sed below, were taken to permit evaluation of the performance of the ceiling, ceiling baffles, and screens. The test area was the middle of the space where reflections from boundary walls were expected to be small. By way of confirmation, propagation measurements before and after installation of absorbing panels on the inner core walls showed no difference.

It should be noted that this study is circumscribed by the propertie s of the installed ceiling, a fis sured mineral fibreboard having only moderate absorption (measured NRC = 0.45). A more definitive study is planned in another space where it will be possible to change the ceiling board as well as the other parameters.

Four sets of measurements were taken along the same pro-pagation path: with and without ceiling baffle s , and with and without an. ｾｾＮｮｴ･ｲｶ･ｮｩｮｧ  screen. The ceiling panels measured 5 by 2 1/2 ft and were installed in metal T-bars. The baffles were made of glass fibre covered with a white porous cloth, measured 8 ft by 8 in. by

2

1/

2 in., and were suspended 1 -Ln , below the ceiling from the T -bars. The arrangement is shown in Figure 23. The intervening screen consisted of glass fibre batts in a frame covered with open-weave cloth. The top and bottom of the screen were 5 it and 6 in. ,

respectively, from the ground. The screens were 6 ft wide and about 3 in. thick, either flat or gently curved. The results of the propagation measurements are shown in Figure 24, and are replotted in Figure 25 to draw attention to the effects of the screens.

-12-Measurements taken from point to point within the space, with no intervening screens or b aff Ie s , show that the absorption is adequately high, giving a decrease in sound intensity of about

5dB(A) with doubling of distance. By comparison, attenuation in free space would be 6dB per doubling of distance. The same

serie s of measurements rnacie after the installation of ceiling baffles shows that this b affIe configuration produced no significant change in the rate of sound a tt e nu ati o n ,

Another se rie s of propagation rrie a su r erne nt s , using the same apparatus, was made as an adjunct to the subjective articula-tion measurements. In this series the loudspeaker was pointed directly at each of the several work positions used for articulation tests and the sound levels noted. Following installation of the ceiling baffles the series wa.s repeated and the two sets of results are compared in Figure 26, again demonstrating the limited importance of the ceiling baffles in this configuration.

The significance of trii s information needs careful con-sideration. Othe r tests h ave shown that baffle devices fixed to the ceiling at sufficiently cIc s e spacing can reduce sound propaga-tion in an open-plan offi c e , at least beyond a certain minimum

range. The area of interest is within a lO-ft :;:oange of any speaker position because it is in this zone that speech privacy and annoyance problems exist. At greater distance s the signal will normally

have sufficiently decreased in intensity to be little more than part of the general noise, and this will even have an advantageous effect on the level of naturally generated rn a s k irig noise. The ineffectiveness of the baffle s, as installed, is probably due to their configuration. Along the propagation path the baffle centre s were lOft apart and the degree of baffle interaction wit h sound waves was understandably

small. It should be added, however, that many favourable comments were received about reduction of glare from £luore scent lighting fixtures.

The effects of free - standing screens on the propagation of sound in the test office may be seen in Figure 26. Screens should not be regarded as a panacea for ac ou st.i c ills in a landscaped office, but with careful use and attention to other important parameters, they can be very valuable in acoustically separating adjacent work

stations.

When considering the performance of a screen, the importance of the ceiling material emerges. Sound, at the important speech

by direct transmis sion through the screen, by diffraction around the screen,

by reflection from the ceiling,

by reflection from vertical flanking surfaces.

For moderately well constructed screens one can expect to achieve attenuations of 15 to 25dB for sound travelling through them. For a given size of screen the amount of diffraction is fix.ed and generally, is not the rn o st important limiting factor at the important speech

frequencies.

The most important paths involve reflections from horizontal or vertical flanking surface s , In the absence of vertical flanking

surfaces, reflection from the ceiling plays the dominant role in determining the performance of a screen. For this reason much attention is given to the absorption of the ceiling material in an open-plan office. More specifically the absorption coefficient for sound wave s incident at angle s around 50 0

is critically related to the reflected sound signal received at nearby work stations. As this latter property varies from material to material, preferences have been expressed for use of some materials over others.

For a screen positioned 6 it from the source, typical

insertion losses are 5dB(A), corresponding to a reduction in articula-tion index of O. 17. For a given speaker-listener distance, the inser-tion loss is greatest when the screen is close to the speaker or

listene r ,

A subsidiary experiment, the results of which are not shown here, indicates that for any separation of 9 ft or greater the insertion loss achieved by positioning a screen 3 ft from the listener is fairly constant at about 7dB(A), co r re sponding to a reduction in articulation index of 0.23. Such changes in AI can alter a situation from "marginal" to " c o nfidential privacy" or from "fa.i r communication" to "normal privacy. II

It should be kept in mind that these results were obtained in an office space with a fairly high density of furniture, including

sound-absorbing screens. Measurements were made along a fortuitously-occuring aisle and it is possible that the presence of the furniture on either side of the aisle interfered slightly with the propagation measurements.

Conclusion

The major problem in a landscaped office is the annoyance caused by intrusive sounds, particularly speech. There is no doubt

-14-that  the  installation  of  a  background noise  system  can  effectively  improve  the  isolation  between  neighbouring  work  places.  It  appears  from this  study,  however,  that  unles s  workers  have  a  strong  need  for  privacy background  noise  is  unacceptable.  Other  workers, 

having  a  stronger  need for  privacy,  show  a  different  reaction. 

When  there  is  a  strong  need  for  privacy  or freedom  from  distraction,  one  might  reasonably  que stion the  open­plan  approach. 

This  study  illuminate s  an  obvious  fact;  that  people  do  not  gladly  accept  added  noise  in  their  lives.  Given the  choice,  they  prefer to  have  quietness.  With this  in mind,  it  seems  that  any  open  office  design  that  requires  the  use  of  high  masking  noise  levels  to  guarantee  privacy  should  be  carefully  considered  and  other  possible  approaches  looked  into.  It  is  also  important  in  making  such  an  assessment that the  desire  for  visual privacy not  be  confused  with  the  need for  acoustical  privacy.  There  will,  naturally,  be  intermediate  cases where  absolute  privacy  is  not  essential  and  a  masking  noise  system  in  an  open  office  can  provide  the  answer. 

In the  environment  studied  a  moderate  level  of masking  noise  is  provided  by  mechanical  equipment.  Although the  character  of  the  mechanical  equipment  noise  is  not  ideal,  it  has  the  advantage  of  being  as sociated with  a  de sirable  attribute  of  the  space  (air  conditioning)  and  it  provide s  the  modicum  of  masking  needed to  make  the  space  tolerable.  Except for  real  privacy  needs  this  seems  to  be  sufficient.  The  physical measurements  were  in  accord  with  expectations,  although the  ceiling  baffles  were  not  very  effective,  probably  due  in  part to  the  spacing}depth  relation used. 

Acoustic  screens  are  most  valuable  for  the  separation  of  adjacent  workspaces  and,  used  intelligently,  can  give  attenuations  of  5­7dB(A),  a  value  limited  primarily by the  properties  of  the  ceiling.  To  avoid  decreasing  the  effect  of  screens,  surfaces  above  desk  level  should  be  absorptive,  where  practicable,  to  eliminate  reflected  sound.  The  value  of  screens  is further  enhanced  by the  fact  that  they  introduce  absorption  into  a  space  and  provide  visual  privacy. 

-

REFERENCES 

­15-1.   Abrahams,  M. L.  Environmental Space  Planning  and  Design  ­ ESPD.  ｏｰｴｩｭｵｭＬｾＬ  No.3,  1971. 

2.   Lappat ,  A.  Criticisms  of  Officescapes,  Bauen und  Wohen  25,  No.  I,  pl ,  Jan.  1971,  (In German,  will  appear  as  National  Science  Library  Technical  Translation Note). 

3.   Lorenzen,  H. J.  and  Jaeger,  D.  The  Office  Landscape.  Contract  Magazine,  January,  1968. 

4.   Hegvold,  L. W.  Acoustical Design  of  Open­Planned  Offices  National  Research Council  of  Canada,  Division  of  Building  Research,  CBD  139,  July  1971. 

5.   Beranek,  L. L.,  Blazier,  W. E.,  and  Figwer,  J. J.  Preferred  Noise  Criterion  Curves  and  their Application  to  Rooms,  J.  Ac ou st ,  Soc.  Am.e r ,  ｾＬ   No.5 (Part I),  1971,  P  1223.  6.   Hegvold,  L. W.  Experimental Masking  Noise  Installation 

in  an  Open­Planned  Office,  National  Research Council  of  Canada,  Division  of  Building  Research,  Technical Note  No.  563. 

7.   Hamme,  R. N.  and  Huggins,  D. N.  The  Problem  of  Acoustical  Specifications  for  Office  Landscape  Ceilings,  Paper Dll,  83rd  meeting  of  Acoustical Society  of  America,  April  1972. 

8.   American  National  Standard Methods  for  the  Calculation  of the  Articulation Index,  ANSI S3. 5,  1969. 

9.   Griffiths,  J. D.  Rhyming  Minimal  Contrasts:  A  Simplified  Diagnostic  Articulation  Test,  J.  Acou st ,  Soc.  Arne r ,  42,  No.  I,  1967,  P 236. 

TABLE  I  TEST  SCHEDULE 

MASKING  NOISE  CONDITION 

DATE  AM 

PM 

Wed  Oct  13  A3  B3 

Thurs  Oct  14  B3  A3 

Fri Oct  15  A3  ambient 

Man  Oct  18  ambient  A3 

Tues  Oct  19  A3  A2 

Wed  Oct  20  A2  B2 

Thurs  Oct  21  B3  BZ 

Fri Oct  22  B2  B3 

Man  Oct  25  B2  Bl 

Tues  Oct  26  B2  C2 

Wed  Oct  27  B2  ambient 

Thurs  Oct  28  C3 

Cl

Fri Oct  29  B2  B3 

Man  Nov  1  till 

Fri  Nov  5  B4  B4 

Man  Nov  8  till 

Fri Nov 12  ambient  ambient 

Thurs  Nov  18  ambient  C4 

TABLE  U

SUBJECT  GROUP  CLASSIFICATION  

POSITION  C LASSIFICA T ION  

Director.  Const ,  Branch  Director's  Sec'y 

Chief  Engineer  Chief  Eng's.  Sec'y  Chief Architect  Chief Arch. 's Sec'y 

Chief Management  Support  Chief Man.  Supp. 's Sec'y  Arch.  Contracts  Officer  Stenographer 

Chief.  Contracts  Division  Chief.  Con.' s Se cfy

Typist 

Authorities  Officer  Contracts  Clerk 

Eng.  Contracts  Officer  Special Contracts  Officer  Contracts  Clerk 

Contracts  Clerk  Financial  Officer  Fin.  Off. IS Clerk 

Fin.  Off.'s  Clerk  Chief Admin.  Services  Chief Admin.  Se r l s ,  Sec'y  Chief Program  Co­Ord.  Mail  Clerk  Purchasing  Clerk  Program Co­Ord.  Officer  Personnel Clerk  Records  Clerk  Records  Clerk 

Elect.  Power  EO.  Design  Off.   Elect.  Power  Design  Off.   Stenographer  

Elect.  Power  EO.  Design  Eng.   Chief  Elect.  Engineering  

Eng. 

ric

Elect.  Power  EO.  Design   Elect.  Power Systems  Eng.  

Elect.  Power Systems  Eng.   Elect.  Power  Systems  Eng.  

SX2  ST6  EN2  CR4  ARB CR3  AS7  ST5  AS4  CR4  AS6  ST4  ST3  ASl  CR3  AS4  AS4  CR5  CR3  FII  CR3  CR2  AS4  ST4  AS4  CR3  CR4  AS2  CR5  CR3  CR2  EG9  EG6  ST3  EN3  EN5  EN4  EN2  EN4  ENB

TABLE  ill  

CALCULATED  AI  IN  EACH  POSITION  FOR   ALL  TEST  CONDITIONS  USING  IDEALIZED  VOICE  

MASKING  CONDITIONS  TEST 

POSITION  A2  B2  Bl  A3  B3  A4  B4  AMBIENT 

Dingwall  _{• 52  • 62}  .55  _• 61  .69  .66  .74  _{·  79}  Coates  _{• 19  • 26}  0.2  _{• 28  • 38}  ·  28  ·  35  • 37  n Pos  X  _{• 31  • 37  • 3  ·  36}  .48  _{·  39}  .46  _{·  50}  Butler  _{• 4  • 47  • 4}  .45  .56  .47  _• 59  ·  62  n Pos  Y  .17  .24  _• 18  .29  • 38  ·  26  • 35  • 50  Somers  .43  _• 51  .45  .44  .54  .48  _• 58  .74  Peach  • 36  .46  • 4  .44  ·  58  ·  47  ·  55  ·  71  Round  _• 22  .28  _{• 21  ·  29  ·  39}  .27  ·  38  .48  McManus  _{• 22  • 3}  .24  .26  • 37  ·  27  ·  21  • 30  Papineau  _• 14  .17  _• 12  • 18  ·  25  ·  17  • 21  ·  32  Anderson  _{• 23  • 25}  0.2  _{·  32  • 38  • 27  • 34  ·  35}  Austin  _{• 1  • 16} 

.11 

• 18  • 27  • 17  ·  23  • 25  McGillivr ay .09  • 13  .09  _• 18  ·  24  ·  17  ·  22  ·  18 

MEAN  SCORE  ON  RHYMING   TEST  POSITION  Dingwall  Coates  Position X  Butler  Position  Y  Somers  Peach  Round  McManus  Papineau  Anderson  Austin  McGillivray  A2  69.5   78   79.5  86   88   68.75  70. 5   40.2  62.7  50.7  36.7 

TEST  FOR  60dB(A)  PEAK  VOICE  LEVEL 

MASKING  B2  69.25  79.7  79.2  91. 5   93   75. 5   73.25  65   57.7  CONDITION  A3  66.5  77.7  77.5  93.5  79.7   76   76   69.7  AMBIENT   97   76   81. 7   84.2  94.2   93   82.25  83. 5   68.6  65.7  71. 5   64.0 

TABLE  V  

MEAN  SCORE  ON  RHYMING  TEST  FOR  55dB(A)   PEAK  VOICE  LEVEL  

MASKING  CONDITIONS  TEST 

POSITION  B2  B1  B3  A4  B4  AMBIENT 

Dingwall  90  77.7  83.25  Coates  45  30.7  58.75  29.75  23.7  55.5  Position  X  53.7  50.25  65  44.7  46.6  66.2  Butler  57.7  48  67.5  44.8  72.7  Position  Y  34.5  43.1  67.0  Somers  75.4  66  74.6  71. 7  85.5  Peach  84.8  71  88  77.0  70.6  90.7  Round  55  43  59.5  43.75  45.4  65.7  McManus  50.5  33.7  64.75  41. 75  34.3  43.5  Papineau  20  43.5  Anderson  46.7  32.5  52.2  Austin  31. 7  33.2  51. 5  McGillivray  20 

TABLE  VI 

SPEECH  PRIVACY  FOR  VARIOUS  POSITIONS 

NOISE  LEVEL 

POSITION 

dB(A) 

FRONT 

40 

" 

40 

47

"

47

"

50

"

50

"

SIDE  

40 

40

"

47

"

47

"

50

"

50

"

BEHIND  

40 

40

"

47

"

47

"

50

"

"  

50 

12  FT  FROM  TALKER 

PERCENTAGE 

SCREEN 

SCORE 

NO 

YES 

NO 

YES 

NO 

YES 

100 

86 

86 

63 

71 

52 

NO 

YES 

NO 

YES 

NO 

YES 

90 

70 

71 

50 

63 

39 

NO 

YES 

NO 

YES 

NO 

YES 

75 

57 

57 

29 

42 

20 

ARTICULATION 

INDEX 

PRIVACY 

CONDITION 

.88 

.65 

.65 

.45 

· 55 

.35 

NIL 

NIL 

NIL 

VERY  POOR 

NIL 

VERY  POOR 

.75 

.55 

· 55 

.32 

.45 

.22 

NIL 

NIL 

NIL 

VERY  POOR 

VERY  POOR 

POOR 

.58 

.35 

· 35 

.12 

.25 

.02 

NIL 

VERY  POOR 

VERY  POOR 

ACCEPTABLE 

POOR 

EXCELLENT 

o

CJ

BWALL

TALKER

D

PAPINEAU

､ｊｾｭｸｸｘｘｘｌ

 

ｲｃｔｾｓｉ  

(]l 

I .

ﾷ｢ｔｾｒｉ

 

SOMERS

o

PEACH \

ｾｾＯ

 

I

I '

ｾ  

FIGURE 1 TOWER C 4th FLOOR LAYOUT

•

• •

60 

W  e::::  co  '"0

50

Z

_•

e:::: 

_•

•

. . J «   Wco  >0 We::::  ..J 

40

U ｃｬｾ   Z « N   co 0 0 

30 

wO  > . « 0   I

-u 

20

0 I Cl e::::  :J:  I

-1000 

10000

100 

FREQUENCY IN HERTZ

FIGURE 2a

MASKING SPECTRUM 81 (DOTS) COMPARED

WIT HAM 8 lEN T SO UND (S 0 LID LIN E) 42 d8 (A )

6 

a __...

｟ＭＮＮＮＮＮＭＭＭＬＭＭＭｾＭ｟ｲＭＭＭＭｲＭＭＭＭＭｲＭＭＭＭＭＭＬｾＭＭＬ  

LU 

0:::  Z  50 

_•

•

_•

...Jo:::

LU« 

>CO 

ｾ ｾ  

40 

oU  ｚｾ  

« 

CON  0 

₃₀

LUo 

> 0

.

« 

1­ 0  U  

0 

_,

20 

• 

0  0::: 

:c 

I

-10000

1000

100  FREQUENCY IN HERTZ

FIGURE 2b

MASKING SPECTRUM B2 (DOTS) COMPARED

WITH AMBIENT SOUND (SOLID LINE) 42dB(A)

ｂｒｾｾＹＰ  -3

co  

•

"1J  

_•

Z 

₅₀ 

•

_•

...J  a:::: 

LU« 

>co 

LUO 

...Ja:::: 

40 

oU  ｚｾ  

« 

co  N  0 

₃₀

LUO 

> 0  

«

. 

0 1-U 

0 

20 

I Cl

•

a:::: 

­

:I:  FREQUENCY IN HERTZ

FIGURE 2c

MASKING SPECTRUM B3 (DOTS) COMPARED

WITH AMBIENT SOUND (SOLID LINE) 42dB(A)

LU 

60 

0::: 

I

•  • 

60

• 

Z  50 

• 

­Ie::::  •

LU« 

>co

LUO 

­I  e:::: 

40 

oU  ｚｾ  

«

_coN  o  30  LUO ＾ｾ

«0 

I -U 

o

20 

I

o 

e::::  FREQUENCY IN HERTZ

FIGURE 2d

MASKING SPECTRUM B4 (DOTS) COMPARED WITH

AMBIENT SOUND (SOLID LINE) 42 dB (A)

1 00

1000 

10000 

N o  o  o  o 

60

LLl 0::: Z  50  0::: --14: LLla:l

ｾｏ

 

--I 0::: 40 

u 

0 -Z::f 4: a:l 30  LLl

>

4: I -U 

o 

20 

I 

o 

0:::  :c  I -1 00 

1000 

FREQUENCY  IN  HERTZ 

FIGURE  2e 

BBN  RECOMMEN  DED  RANGE  I N  THE  NO  I SE 

SPECTRUM  IN  AN  OPEN  PLAN  OFFICE 

10000  

ACOUSTIC SURVEY OF TOWER C National  Research  Council  of  Canada  

Division  of  Building  Research   DATE  1  _

PLEASE CHECK ONE OF THE FOLLOWING: much  prefer  morning .•..•....••••.. 

D

slightly  prefer  morning ...••.•. 

D

no  preference ...•....•.•..•.•.•• 

D

slightly  prefer  afternoon .••.•...•. 

D

much  prefer  afternoon .•.•..•....•.• 

D

OTHER COMMENTS

1 0 45 dB(A) 48 dB(A) 26/10, 29CARDS 5dB/OCT 20 Vl LU  Vl Z 

o 

Q.. Vl LU  ｾ   LL. 1 0

o 

ｾ   LU  co  :E ::>  Z  ｾ   LU  LL. LU  Q.. I U ::>  :E .,J::t::. ｾ   LU  LL. LU  ｾ   Q..

>-...J  I -I 0 

­

...J  Vl LU  U Z  LU  ｾ   LU  LL. LU  ｾ   Q.. 0 

z

ｾ   LU  LL. LU  ｾ   Q..

>-...J  l -I 0  -...J  Vl ｾ   LU  LL. LU  ｾ   Q.. I u ::>  :E CONDITION CONDITION ONE TWO KEY

FIGURE 4

1 0 10 10 48dB(A) 13/10, 35 51 dB(A) CARDS 48 dB(A) 14/10, 51 d B(A) 33 CARDS 48 dB(A) 51 dB(A) MEAN -7dB/OCT

FI GURE 5

1 0 AMBIENT 48dB(A) 27/10, 25 CARDS -5dB/OCT

FIGURE 6

BR ...ｾｾｏ  -8

1 0 10 1 0

AMBIENT AMBIENT 51 dB(A) AMBIENT 51 dB(A)

7dB/OCT. MEAN 15/10, 28 CARDS 18/10, 33 CARDS

FIGURE 7

51dB(A) 1 0 AMBIENT 45dB(A) D IFF USE RS 18/11,25

FIGURE

10 1 0

AMBIENT 48 dB(A) AMBIENT 48dB(A) 8/1 1 1/1 1 8/1 1 1 /1 1

FIGURE lOa

FIGURE lOb

1 0

LOWER dB (A) HIGHER MEAN

FIGURE 9

0 FF CARDS

8

ＶｾＧｬＧｬｾｯＭＹ  

10  1 0 5 dB/OCT. 7 dB/OCT. 28/10,  22  CARDS  45 dB (A)

FIGURE 11

10  1 0 5 dB/OCT . 7 dB/OCT. 5dB 7dB 5dB 7 dB 5 dB 7dB 21/10,  28  CARDS  22/10,  23  CARDS  29/10,  25 CARDS  MEAN 

48 d B(A)

FIGURE

12

5dB/OCT. 7 dB/OCT. 20/10,  29CARDS  51  dB (A)

FIGURE 13

1 0 1 0 1 0

5 dB/OCT. 7dB/OCT. 5 dB/OCT. 7 dB/OCT.

19/10,  29  CARDS  MEAN 

1 0 5 dB/OCT. 3 dB/OCT. 48 dB(A) 25/10, 26 CARDS

FIGURE 14

80 . . . . - - - - . - - - , . - - - , - - - - r - - - r - - r - - -...- - - - , - - , - - , Z  70  ­10::: 

w« 

> d l   wO  - I 0:::  60 ｃｬｾ   ｚｾ   « _N dl  o  50 w O  ＾ｾ «0  t-u

o

40 I Cl 0::: :::c t-30 PE A KS R. M. S . 1000 10000 FREQUENCY IN HERTZ

SPEECH AT 1 METRE, MALE VOICES

LONG TERM

r. m. s.

=

65 dB)

BR"''''fO-I\ 100

FIGURE 15

IDEALIZED

( OVERALL

1.0 

o. 

8 

· 6 

«

· 4 

· 2 

NO.4 SPECTRUM

I , I I I I I I I ,

o

,-­17

­13 

­9

­5 

­1 

3 

7 

11 

15

19 

23 

SIGNAL TO NOISE RATIO, dB

FIGURE 16a

AI VS SIN FOR IDEAL SPEECH SPECTRUM

1. 0

I i i i i i I i i i "'"""

· 8 

· 6 

«

· 4 

· 2 

0 

,

­13 

­9 

­5 

-1

3 

,

7

11

15 

19 

NO.

NO.3 SPECTRUM

' 

,.,...-

' . . . - . = : ' , I i

­17 

i '

SIGNAL TO NOISE RATIO, dB

i

23 

I

FIGURE 16b

AI VS SIN FOR IDEAL SPEECH

SPEECH LEVEL AT 3FT TAKEN AS 65 dB LONG TERM

r.  m.  s. 

NOISE LEVELS IN dB A

O.  9 

O.  8 

O.  7 

x ｾ   0.6 

z 

z 

o

;::  0.5 

«

....J  :=l U l -e::: 

0.4 

« 

O.  3 

0.2 

O.  1 

o 

HORIZONTAL  SPEAKER  ORIENTATION  

FIGURE  17  AI  VS.  SPEAKER  ORIENTATION 

SR ,,"'.0 -1*

­­­­­­­­­­­­­­­­­­­­­

NOISE CONDITION u.J  "" 

o 

u V"I Cl  u.J  ｾ   0.6  ....J « ::E

""

o 

z 

z 

0.4  « u.J  ::E u.J 

"" 

o 

1.0  u V"I Cl  u.J  N ....J 0.8  « ::E

"" 

o 

z 

Z  0.6  « u.J  ::E Cl  u.J  I-U 0.4  u.J  Q..  X u.J  0.2  B4 A4 Bl B2 B3 0.8  \ /

"

"

/ \. • SPEAKER 1 0SPEAKER 2 (0  ) 

• 

_ｾ   -, \.  0 ­ ­ 0 ­ ­ 0   ­, _{• _ .} ... _...0­­0 _jO

--.-...""""

_,

. / <,

.

_. 0  /0

-'...

-,

­. 

'0"'/"""-. : , •  . /   ... <, / / • GRIFFITHS •

o 

ANSI (b) (60) (60) (55) (60) (55) (55) (55) (55) (55) (55) VOICE LEVEL FIGURE 18

EXPERIMENTAL AND EXPECTED MEAN NORMALIZED

•

_• 

•

••

•

•

• •

••

•

•

•• • •

• •

•

•••

_{•  •} 

• ••• 

• • • •

• 

• 

• 

•

•

.­

Vl 

100 

L.U

.-C>  Z ｾ  

80

>-I e:::::  Z

o 

L.U

60 

e::::: 

o 

u 

Vl  L.U C> 

40 

«

.-z

L.U U

ｾ  

20

c,

z

«

L.U ｾ  

o

•

•

: . .

.

ｖＢｾＺ

·

ｾ  

. /

...

ｾＮ

 

".

.

..

ｾ

 

.

.

ｾ

..

•

•

ｾＮＯ

.

...-

..

r

O. 82

..

ｾＮＧＮ

 

•

O. 1

O. 2

0.3

0.4

0.5

0.6

0.7

O. 8

0.9

LO

CALCULATED ARTICULATION INDEX

FIGURE 19

MEAN SCORE ON RHYMING TEST VS CALCULATED A I

•

_•

•

••

•

_•

• •

•

•

I I I I I I

•

l

-100 

• • 

ｾ

 

.

•

..

Vl

•

•

•

•

•

•

l..LJ l - _ｾ  

-c.:>

₈₀ 

•

•

•

•

-

_•

ｾ  

•

• • • •

• •

:z 

•

. .

»:

­

...

.

>-:::I: ｾ  

60

•

-:z

ｾＮ   0

•

l..LJ 0 u ｾ  

40

/!

.

•

•

•

•

•

Vl -l..LJ I ｾＭ c.:> ｾ  

•

:z

I -

ＲＰｾ

 

_•

_•

l..LJ U ｾ   l..LJ

0-a

O. 1

0.2

O. 3

0.4

O. 5

0.6

O. 7

O. 8

O. 9

1.0

ARTICULATION INDEX

FIGURE

20

BEST FIT OF TEST DATA TO GRIFFITHS' RESULTS

ｂｾｌＫＹＹｃ｜Ｍ｜ＢＢ  

• •

• 

_•

• •

••

• 

_•

_•

•

• 

_• 

• 

••

_• 

•

_•

•

•

•

•

0-a  00 a 

• 

r-

_.

• 

a 

• 

\,Q

• 

•

•

•

a 

•

•

•  •  

•

Lt'\

•

a

•

••

ｾ  

...

_• 

..

•

• 

_•

•

­er

• 

a

• 

••

• 

•

•

•

C""

•

a 

•

N a 

•

•

l"""'4 a 

•

a  a a  a a  a  00 N l"""'4

1531

ｾｎｉｗａｈｾ  

NO

ＳｾｏｊＵ   Ｓｾ｜ｦＱｎＳｊｾＳ､   X LJ.J  0  Z  Z  0 I -<t:  ­­­J  =>  U  I ­ c::::  <t:  <t:  I -ｾ <t:  0  _Q 

..

0-:I"  Ii V')  Z  b  <t:  0  LJ.J  u...  0  0  :2:  0  l -<t:  l -<t:  0  l ­ V')  LJ.J  I ­ u...  0  l ­ u...  l ­ V')  LJ.J  co  l"""'4  N  LJ.J  c::::  =>  <.:>  u... 

POWER

PINK NOISE

AMPLIFIER

GENERATOR

d

..

..

ｾ

[L]

V

EV 8"

SOUND LEVEL

_LOUDSPEAKER

METER

₃

1

_{FROM FLOOR}

FIGURE 22

APPARATUSUS EDIN PRO PAGAT ION MEA SUR EME NTS

BAFFLE

FLUORESCENT

LIGHTING

FIGURE 23 SCHEMATIC Of CEILING LAYOUT

nn. "'99"  ­ 19 

o  

­10 

­ 20 

<..=:> 

z 

0

« 

l.L.J  0::: 

-('t"\

0 

0  I­-l.L.J 

> 

I­­

­10

« 

...J  l.L.J  0::: 

« 

ｾ  

­ 20 

"0  Z  Z  0 

I­-« 

ｾ  

0 

z 

UJ 

I­-« 

­10 

­ 20 

f...-­

f...-­

­

­

a

I 

I 

I

I

I 

ca 

aNO  BAFFLES 

·BAFFLES

• 

a

• 

• 

ｾ  

•

NO  SCREEN 

a 

I I  I I I   I I I I  I  

a 

ｾ  

• 

_e 

• 

_• 

SCREEN  AT 

3

1  ｾ   I I I I 

_a

1 I  1 1 

• 

a 

a 

•

• 

_a

• 

•

• 

0

• 

SCREEN  AT 

6

1 

a

•

I 

I 

I

I

I

3

1 

6

1 

12

1 

₂₄

1  481 

DISTANCE  FROM  LOUDSPEAKER,  FT 

FIGURE  24 

PROPAGATION  MEASUREMENTS  SHOWING  BAFFLE 

­

­ ­ ­

­

­

­

­

EFFECTS  

...

0

I

0 

,.--10

0 

•

0  ...J  0 L..LJ 

>

•

L..LJ  0 ...J 

- 20

0

-

•

C'f'\

NO BAFFLES

a

₀ I--w

•

j

«

>

I I--0 

NO SC REEN

...J 

•

SCREEN AT 3

1  L..LJ  ₀

6

1 0::: 

SCREEN AT

0

0 

«

0  co  "'C ₀

z 

_-10

0  0 

z 

• 

0 ₀ 0 

a

0 I--

•

• 

₀ 

«

₀

_•

::J 

- 20

• 

0

z

_• 

0 L..LJ

I--«

₀

-30

•

BAFFLES INSTALLED

­40 

3' 

6

1 

12

I 

24

I 

48

I 

DI STANCE FROM LOU DSPEAKER, FT

FI GURE 25

PROPAGATION MEASUREMENTS SHOWING SCREEN

EFFECTS

-10

z 

o 

20  30 40

o

BEFORE BAFFLE INSTALLATION 0 MEAN = 15.4 • AFTER BAFFLE INSTALLATION

•

MEAN - 16.0

>-X

>-

<J::  ...J c>::: Z  ...J _{Z Z}  Vl _:::J

>

:::J 0 <J::  _{Vl 0 0 Vl} <J:: 

­

_Vl ....J c>::: 0 Z  w  ...J Z ｾ   w  c>::: c>::: I t­ w t­ Z t­ w  Z <J::  L.LJ 0

_­

...J

­

_U t-<J:: Vl Vl ｾ   :::J ｾ   _0... 0 0 Vl Z t- _<J:: u u Z 0 ₀ :::J 0 0 w 0 <J:: :::J 0 U 0... co 0... Vl 0... c>::: ｾ   0... ｾ   <J:: <J::

FIG UR

E 26

PROPAGATION MEASUREMENTS TO THE TEST POSITIONS

Resume  of  introductory talk given to  the  participants  in the  masking-noise evaluation te st.

"As you are no doubt aware, your new accommodation js pro-vided with a background sound system which is presently in 9f'>eration. It is believed that such a system, when properly tuned and used in an open office environment, will improve the speech privacy condition, and reduce annoyance from intelligible speech and machine noise, thus providing improved working conditions.

We propose to carry out a series of tests to establish the most desirable operating condition for the system. The tests will be in two parts.

For the fir st part of the te st, you will be expo sed each day to two different masking noise conditions, one in the morning and one in the afternoon, and you will be asked to mark your preference on a card which will be provided. These cards will be picked up and analysed each day. There is a space on the card where you can make comments if you feel like doing so. In so doing we hope to arrive at the most satisfactory operating condition for the system.

The second part of the test will not involve the people working in the area; a different test group will be used. Articulation tests, to measure the effectiveness of the masking system for improving

speach privacy will be held each day during the lunch hour. It is for this reason that you have been asked to vacate the office during lunchtime. "

l