Dept. for Speech, Music and Hearing

Quarterly Progress andStatus Report

Sound levels and long-timespectra recorded within the

symphony orchestra and riskcriteria for hearing loss

Jansson, E. V. and Karlsson, K.

journal: STL-QPSRvolume: 23number: 1year: 1982pages: 031-048

http://www.speech.kth.se/qpsr

sents the f i r s t p a r t of an investigation of the sound environment and hearing among the members of symphony orchestras. The second part, the

r e s u l t s of an audiological examination of t he musicians, is reported

elsewhere (Karlsson e t al). I I

1. Risk cr i ter ia

]Loud noise may cause hearing loss. The loss depends on three main

factors: the leve l of the noise, i ts spec t ra l content, and i ts dura-

tion. Risk c r i t e r i a fo r in jur ious noise i n working places have been

worked out and standardized (Swedish Standard, SEN 59 01 11, 1972). me leve l of the noise is measured i n dB(A), sound pressure level , - t h a t

is, with a f i l t e r designed t o approximate the ear's s e n s i t i v i t y at

different frequencies. A t widely varying noise levels, the to ta l dose

should be measured - the equivalent sound level, Leq (the time average

of the sound intensity converted t o sound pressure level i n dB). In the

equivalent sound level, loud sound components are given greater weight.

The Leq should be measured during - or should be converted to - a r e p r e

sentative 40-hour work week. The r isk threshold for noise injuries is

s e t a t 85 dB (A), Leq.

A t sound leve ls exceeding 85 ~ B ( A ) , moreover, the sound leve l i n

different frequency bands should be recorded, i.e. the spectral proper-

t i e s of the sound. In t h i s risk-case the leve ls i n the d i f f e r en t fre-

quency bands should be weighted against their duration. The risk-limits

may best be described in this case i n the form of several curves i n the

same diagram. Each curve then describes t he leve l t h a t may not be

exceeded for more than a specific time.

A t very high sounds levels (exceeding 140 dB(A)), also with sounds

of very brief duration, but longer than 20 msec (impulse sounds), there

is an immediate risk of hearing injuries. In our analysis, we should

therefore record such r isky short-time leve ls where they occur. N o

further risk c r i t e r ia have yet been established for these sounds i n the

Swedish standards.

2. Analysis material

In t h i s study we recorded sound pressure leve ls on the podium

during three concerts given by the Radio Symphony Orchestra and Stock-

holm Philharmonic Orchestra, as well as in the orchestra pit during two

ballet performances with music by the Opera Orchestra. The same proce-

dure was used to record the analysis material for the three orchestras.

From the orchestras, programs were received concerning rehearsals,

concerts, and performances, from which we made a selection of "heavy"

music. Moreover, we chose to make the recordings during a complete

concert or performance since, in these cases, the music follows a given

pattern without interruptions or replays (see Table 11-A-I). During a

rehearsal we selected three or four microphone positions, two (or one)

of which, in our opinion, were "normal" positions, and two of which , in our opinion and that of the orchestra members, were "exposed" positions.

The microphones were placed beside the musician at ear level, at a

distance of 0.3-0.5 m. \

The recordings were made with measuring-microphones (B&K 1/2-inch)

on two of the own tape recorders of the orchestra. Signals for absolute

calibration and checking of the frequency response were also recorded.

With help of the calibration signals the recording levels of the tape

recorder were adjusted so as to record high sound levels without audible

distortion. This resulted in a dynamic range from 124 to about 65 dB

SPL .

3. Analysis methods

The sound levels and sound spectra within the symphony orchestra

depend on a multitude of variables, e.g., type of music, composition of

the orchestra, position in the orchestra, design of the orchestra podium

and so on. An example of how the sound level may vary over a period of

time is shown in Fig. 11-A-1. This means that we cannot give a simple

and exact description of the sound. We must therefore give a statistic-

al description. To simulate properties of the hearing the levels should

be measured after a filter with A-weighting, i.e., with a suppression of

high and low frequencies. We then measured the sound levels by means of

a sound-level analyzer (B&K 4426), which produces diagrams, as shown in

Fig. 11-A-2a. From such diagram we can read directly what percentage

of the playing-time the sound level has a particular value. We can also

measure the cumulative distribution, i.e., a curve from which one can

Table 11-A- 1 .

ORCHESTRA : MICROPHONE POSITIONS AT/IN FRONT OF* MUSIC (TYPE) DURATION EXPOSED POSITIONS NORMAL POSITIONS

1 2 3 4 R a d i o Symphony:

1st c o n c e r t -- M o z a r t V i o l i n C o n c e r t o K 2 1 8 ( L ) 2Omin - - V2/Oboe V a / ~ a

S j o s t a k o v i t j Symphony 1 0 (H) 52 " C A / T ~ ~ H n / P e r c VZ/P icc

S t o c k h o l m P h i l - h a r m o n i c :

2nd c o n c e r t -- Rouse1 1 S u i t e i F (H) 15min

S a i n t - S a e n s P i a n o C o n c e r t o 5 (M) 2 7 "

Dvorak Symphony 7 ( H ) 39 "

3 r d c o n c e r t -- R e z n i c e k Uver- t u r e Donna Diana (H) 4min

C h o p i n P i a n o C o n c e r t o 1 (M) 3 7 "

E l g a r Symphony 1 ( H ) 5 1 "

O p e r a O r c h e s t r a :

1st b a l l e t -- M a s s e n e t Manon

A c t 1 ( H ) 39min va** /Tp t imp F 1

2nd b a l l e t -- Haydn f r o m Symfony 1 0 1 v e r y low s o u n d l e v e l , e x c l u d e d f rom a n a l y s i s

Nordheim Ar i a d n e (H) 25min va** /Tp t C B / P ~ ~ C F 1

J a n a c e k CB/Perc S i n f o n i e t t a (H) 26 " II +Tpt II

(TYPE) ( L ) , ( M ) , a n d ( H ) s t a n d f o r " L i g h t " , "Med ium" a n d "Heavy" s o u n d l e v e l s c l a s s i f i e d i n a d v a n c e b y a u t h o r s .

* P o s i t i o n c o d e s : CA s t a n d s f o r C o r A n g l a i s , Bs f o r B a s s o n , T p t f o r T r u m p e t , Hn f o r F r e n c h H o r n , P e r c f o r P e r c u s s i o n , V2 f o r Second V i o l i n , Va f o r V i o l a , F1 f o r F l u t e , P i c c f o r P i c c o l o , T i m p f o r T i m p a n y a n d CB f o r D o u b l e B a s s . c A / T p t m e a n s m i c r o p h o n e a t t h e e a r o f t h e CA-p laye r i n f r o n t o f t h e T p t , e t c .

** M i c r o p h o n e p o s i t i o n e d o u t s i d e p r o t e c t i v e s h i e l d s a r o u n d Va- p l a y e r s .

read directly what percentage of the playing-time the sound level ex- ceeds a particular value (see Fig. II-A-2b). Four sound levels, L90,

L50, L10, and L1, have also been marked i n Fig. II-A-2b, which corre-

sponds to the levels which the music exceeds during 908, 508, lo%, and

18 of the playing time. From the curves in Fig. II-A-2 Leq can be

calculated.

With Leq measurements exceeding 85 &(A) it is specified that the

spectral characteristics of the sound should be specified in octave

bands. We must therefore measure Leq a t various frequencies. Fig. II-

A-3a shows an example with three curves of such measurements. The three

curves represent three different sound levels - approx. fortissimo (ff),

mezzoforte (mf), piano (p) - selected from plottings of the variation of

the sound level with time (cf. Fig. II-A-1). The measurements here were

made with an equipment adapted to the characteristics of the hearing

(1/3 octave f i l t e r s , which for low frequencies have been combined to

correspond to the critical bands of hearing, Bark - see Elenius, 1980).

The r i s k l i m i t s in accordance with the standard norm are converted t o

our measuring method and are shown i n Fig. II-A-3b. From measurements

of the equivalent sound levels and the sound level distributions (cf.

Fig. II-A-2) the percentage of playing-time corresponding to each curve

has been calculated, i.e., the playing-time which gave the Leq for the

corresponding curve.

By means of recorded calibration signals the absolute sound level

was determined. Control measurements were also made of the frequency

response between recording and replaying, in which fairly large devia-

tions were noted in individual cases. The effects of these deviations

were simulated with a f i l ter and it was established that the true levels

with few exceptions f a l l within plus minus 1 d B of the Leq specified

below. The deviations also affect the long-time spectra. However, this

influence did not disturb the relations between risk l i m i t s and the

noise exposures.

4. Results

Some examples of recorded sound levels are given in Fig. 11-A-4.

The risk limit is also plotted for a representative week of noise ex-

posure, i.e., 85 ~B(A) Leq. Thus, we see that the Leq measured for the

"light" music, the Mozart concert, lies 7-10 dB below the risk limit.

We also see that the Leq measured for the "medium-weight" music, the

piano concerto, lies at the risk limit, plus minus 2 dB. The Sinfhet-

ta lies 5-12 dB above the risk limit.

The Leq measured on all the occasions and in all positions are

shown in Table 11-A-11. Leq have also been calculated for all "exposed"

and all "normal" positions combined, i.e., Leq corresponding to all

recordings made in these positions. It should be noted that Leq for

both the "normal" and the "exposed" positions exceed the risk limit.

The table shows that our classification of the musical items (L),

(M), and (H) generally agrees with the sound levels recorded. The

highest levels in an "exposed" position were measured in the Opera

Orchestra where the musicians sit close together. Otherwise, there are

no evident differences between the orchestras.

Taking the level L1 as a measure of the maximal sound levels, we

find that these fall below 124 dB. Also the 0.2% level lies well below

this risk limit (not higher than 3 dB above Ll). We found, analyzing

the time-course (with an effective integration time of 10 msec, as shown

in Fig. 11-A-1) that the sound level from the complete material cnly in

few cases exceeds 120 dB(A) by few decibels, and, thus, remains well

below 140 dB(A) limit.

The L1 level lies about 10 dB above Leq in all cases. Thus, it

should be possible to say, as a rule of thumb, that the maximal sound

level for music is 10-15 dB above Leq.

The Leq which we measured was, in some cases, considerably above 85

~B(A). We shall therefore also determine the levels in the various

frequency bands, i.e., the spectral qualities of the sound.

Fig. 11-A-5 gives a typical example of such measurements for three

dynamical levels, approximately p, mf, ff, with estimated durations for

the music which gave the corresponding long-time spectrum. In the "viola

Table 11-A-11.

1.c q DB(A) AT MUSIC (TYPE) DURATION EXPOSED POSITIONS NORMAL POSITIONS

1 2 3 4

let concert -- Violin Concerto (L) 2Omin - - 77.8 75.1

Symphony (H) 52 " 92.7 92.4 89.5 88.4

2nd concert - Suite in F (H) 15min - 95.8 90.1 88.3

Piano Concerto ( M ) 27 " - 84.8 83.0 84.3

3rd concert - Uverture (H) 4min 92.6 93.1 89.8 89.8

Piano Concerto (M) 37 " 87.0 85.8 83.7 84.7

Symphony (H) 51 " 94.0 93.8 89.0 89.4

let ballet -- Manon Act 1 (H) 39min 93.4 90.6 92.5 -

2 " 4 4 " 94.3 90.0 91.8 - 3 " 2 3 " 98.6 93.2 93.6 -

2nd ballet -- Ariadne (H) 25min 88.8 86.5 87.7

Sinfonietta (HI 26 " 96.7 95.0 89.5

Leq for all music recorded at "exposed" positions 93.1 ~ B ( A ) at "normal" positions 88.9 dB(A)

I 1

POSITION AND MUSIC

Fig. 11-A-4. Recorded sound pressure levels during p e r f o m c e , in "light" music, the Wzart violin concerto ( M I ) in positions V2 and Va , in "middle-weight " music, the Chopin concert (C1) i n positions V2, Va, Pa, and F1, and in "heavy" music, Janacek Sinfonietta (J3) in positions Va, B, and F1. The f i l l ed cir- cles on each vertical bar mrk Leq, the open cir- cles from top t o bottom L1, L10, L50, and L90. The l i m i t for permitted long-tirrre noise dose Ieq 85 &(A) is mrked with a thin horizontal l ine, the broken-dotted l ine mrks noise level of tape.

FREQUENCY ( k H z ) FREQUENCY (kHz)

PITCH (Bark) PITCH (Bark) a) b)

FREQUENCY (kHz)

c) . PITCH (Bark)

Fig. 11-A-5. Recorded long-tirrre spectra from performances. a) within a string position section (Va in

the Radio Symphony Orchestra), b) within a mdwind section (flute in the

Opera Orchestra) , c) in front of trmpet (Cor anglais in the

Radio Syrrp?hony Orchestra) . The top spectral envelope (the "zig-zag" line) corresponds to ff, the middle to mf, and the lowest to p. The mth lines mark the m i - mum allowed levels for exposure tims of mx 5 min, 20 m h , 1 hour, 2 hours, and 5 hours in one day.

Wle 11-A-IIIa. Calculated durations of rrd- and ££-portions which have exceeded spectral r i sk l imi ts (cf. Fig. 11-A-5). Tne p-portions never exceeded any r i sk l i m i t and has thus been excluded.

ORCHESTRA

TOTAL CALCULATED MUSIC DURATION MUSIC "exposed" p o s i t i o n normaln p o s i t i o n

DURATION (mf) ( f f ) (mf) ( f f ) m i n uiin min min min

Radio Symphony 72 CA 52 6 Va 52 8

Stockholm Phi lharmonic 173 B s 114 5 Va 117 36

Opera Orches t r a 157 Va 157 47 F1 157 117

Table 11-A-IIIb. From Table 11-A-IIIa calculated m i m u m allowed playing time per day according to the spectral r i sk limits.

ORCHESTRA M A X I M U M ALLOWED PLAYING TIME

"exposedtt p o s i t i o n ttnormaltt p o s i t i o n min min

Radio Symphony CA 83 (mf) Va >300 ( m f , f f )

Stockholm Philharmonic B s 91 (mf) Va 177 (mf)

Opera Orches t r a Va 67 ( f f ) F1 80 ( f f )

STL-QPSR 1/1982

Table 11-A-IV.a Leq for different carbinations of "light", "mdium", and "heavy"msic (the "medium" porportion msic is the portion mis- sing t o 1 .O) : "normal" position.

Proportion o f "light" music 75 dB(A)

Table 11-A-IV.b Leq for different corbinatians of "light", "mdium", and "heavy" msic (t! 'Imedium" proportion msic is the portion mis- sing t o 1 .O) :- "exposed" position.

Proportion o f "light" music 75 ~ B ( A )

Table 11-A-m.c Level correction in Leq depending on effectivle playing-time per week.

Effective playing time (hours) 2 4 8 12 16 20 24 28 32 Level correction (dB) -13 -10 -7 -5 -4 -3 -2 -1.5 -1

Effective playing time (hours) 36 40 44 4 8 52 56 6 0 Level correction (dB) -0.5 0 +0.4 +0.8 +1.1 +1.5 +1.8

If only the recorded music is played, the "noise-dose", according

to the Leq criterion, has reached the limit after six hours of playing-

time in one week in the "exposed" position. In the "normal" position

one can play for 16 hours before reaching the noise limit.

The sound spectrum is such that the allowed playing time in the

"mrmal" position may be decreased to 7 hours per week in extreme cases.

This means somewhat more than a 50% reduction of the maximum allowed

playing-time in the "normal" position. In the "exposed" position only

minor changes occur. It should be noted that the Leq results refer to

measurement values determined only by the recorded music. The durations

of the spectrum characteristics refer to calculations based on rather

rough quantitation steps. Despite this uncertainty it is probable that

the spectral characteristics of musical sounds may considerably shorten

the permitted exposure time. The spectral characteristics and the aver-

age durations of the mf- and the ff-portions are such that the loud

portion often represent the most "dangerous" part.

Assuming that the effective playing-time is 65% of the total work-

ing time, one can play "heavy" music for about 25 hours a week in the

"normal" position according to the Leq criterion but only for about 10

hours a week in the "exposed" position. The criterion for the spectral

characteristics gives a sharper limitation in the "normal" positions to

about 10 hours a week in extreme cases.

6. Prediction of Leq

Since the various pieces of music have different Leq and different

playing-times, it is hard to estimate the Leq for a longer period of

exposure to noise. Our measurements, however, provide a possibility of

estimating the influence of various factors, such as kinds of music,

effective playing-time, and playing-position.

On the basis of our collected material from three orchestras, we

found that Leq in "light" music is about 75 ~B(A), in "medium-heavy"

music about 85 dB(A), and in "heavy" music about 90 dB(A) in the "nor-

mal" position and about 94 ~B(A) in the "exposed" position.

The dependency of the sound level on the proportions of "light" and

"heavy" music i n the "normal" and the "exposed" positions was calculated

and is shown i n Table II-A-lS7.a and Table II-a-1V.b.

By f i rs t calculating the proportions of "light" music and "heavy"

music in the to ta l playing-time we can read off Leq directly from the

tables. The tables show that just a minor portion "heavy" music suf-

fices to give this music a dominating influence over the "light" music

i n the Leq.

Leq for one working week also depends on the effective playing-

time. This dependency can easily be calculated; it is shown in Table

II-A-V .

It is seen that Leq i s relatively insensitive to variations between

20 and 60 hours a week in the effective playing-time (a halving of the

playing-time means a reduction of the level by 3 d ~ ) . A correction

from a playing-time of 100% to an effective playing-time of 65% (as

estimated by orchestra m e m b e r s ) results in a 2 dB reduction of Leq.

With help of the tables Leq can be estimated. First, we calculate

the effective playing time for "light" music, "heavy" music and the

to ta l playing-time during one week. Then we work out the relative

proportions of "light" and "heavy" music in the total playing-time ard

thus from Table II-A-IV we can obtain the Leq. From Table II-A-V we

obtain the correction in dB for a playing-time that is lager or shorter

than the 40-hour week. This correction is added to the level shown in

Table II-A-V.

Example: If the effective playing-time is 20 hours, including

eight hours of "light" music and four hours of "heavy" music, the pro-

portion of "heavy" music is then 4/20 = 0.2 and that of "light" music

8/20 = 0.4. I n Table II-A-V as we see that the music has a level of 85

&(A) in the "mrmal" position. An effective playing-time of 20 hours

gives a correction of -3 dB. Overall, we thus obtain the following

estimates of Leq:

85 - 3 = 82 &(A) in "normal" position, and

88 - 3 = 85 &(A) in "exposed" position.

High sound levels occur in our analyzed material mainly consisting

of "heavy" music. The sound levels are such that even with moderately

long rehearsal and concert durations the defined risk limits are ex-

ceeded. The limits for "noise doses" in "exposed" positions are reached

after about 10 hours' working-time per week and in "normal" positions

after about 25 hours' working-time per week. The spectral properties of

the music are such that the limits for "noise doses" are exceeded in

both positions after about 10 hours' working-time in a week. Our re-

cordings show no short-time levels exceeding 125 dB(A). Therefore, our

conclusion is that if the risk criteria for hearing injuries caused by

noise also applies to music, measures should be taken to reduce -ure

to noise when 'keavy" music is played.

In a following separate investigation the musicians' hearing was

measured. The results showed that musicians as a group did not have

impaired hearing (Karlsson et al.). Symphony orchestra music therefore

probably does not constitute a risk of a hearing loss, despite the high

sound levels that occur. This means that, at least, a revision of the

defined criteria is needed to predict risks of hearing losses for the

symphony orchestra musicians.

High sound levels do not only mean the risk of hearing loss but

also difficulties in hearing other people and oneself within the orches-

tra. This may be the most serious problem an orchestra member will

encounter.

We wish to thank the orchestras, the recording technicians, and

Bruel & Kjaer Co. for their helpful cooperation.