Using ISO3382 measures to evaluate acoustical conditions in concert halls

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.

International Symposium on Room Acoustics [Proceedings], pp. 1-10, 2004-04-01

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=52a660e5-2156-4999-bf54-bb43c3ef03e4 https://publications-cnrc.canada.ca/fra/voir/objet/?id=52a660e5-2156-4999-bf54-bb43c3ef03e4

NRC Publications Archive

Archives des publications du CNRC

This publication could be one of several versions: author’s original, accepted manuscript or the publisher’s version. / La version de cette publication peut être l’une des suivantes : la version prépublication de l’auteur, la version acceptée du manuscrit ou la version de l’éditeur.

Access and use of this website and the material on it are subject to the Terms and Conditions set forth at Using ISO3382 measures to evaluate acoustical conditions in concert halls

Using ISO3382 measures to evaluate acoustical conditions in concert halls

Bradley, J.S.

NRCC-46884

A version of this document is published in / Une version de ce document se trouve dans : International Symposium on Room Acoustics Design and Science,

Hyogo, Japan, April 11-13, 2004, pp. 1-10

Using ISO3382 measures to evaluate acoustical conditions in concert halls

John S. Bradley

Institute for Research in Construction, National Research Council 1200 Montreal Rd., Ottawa, Canada, K1A 0R6

Email address: john.bradley@nrc.ca

ABSTRACT

Application of the ISO3382 standard can lead to the acquisition of large amounts of data describing conditions in a hall. The data could include the values of a number of measures at 6 or more octave band frequencies and for many combinations of source and receiver location. This paper discusses and gives examples of using this data to find important acoustical features. The amount of data can be reduced by calculating average values over the entire data set or for each sub-area of the hall. Various important spatial variations of acoustical conditions can often be better understood from plots of values versus source-receiver distance. The analysis approach will depend on the purpose of the study, which could be for comparisons with various criteria, for investigations of problems, or to better understand the acoustical properties of the hall. The significance of new measurements can be determined by comparing values: with proposed ideal criteria, with values in well-known halls, or with theoretical predictions. The significance of differences between two values should be considered in terms of published just noticeable differences for particular measures. Separately examining early and late sound levels can be a useful diagnostic tool for better understanding the acoustical properties of halls. KEYWORDS: Early-arriving sound, late-arriving sound, concert hall acoustics measurements

INTRODUCTION

The ISO3382 standard specifies how to measure a number of well-accepted room acoustics parameters and includes guidance concerning, the numbers of source and receiver positions to be used and the calculation details for each parameter. The combination of many different acoustical measures at many frequencies and from many positions in a hall can result in a large amount of data that may at first hide interesting acoustical features. This paper discusses and gives examples of how to focus in on these more interesting, and often more important features.

Many Measures: The ISO3382 includes a number of well-accepted room acoustics measures in the main body of the standard and in appendices. As newer measures become

accepted, these too may be added in future revisions of the standard. However, it is advisable to first focus on the more basic and more generally important parameters. Table 1 lists some acoustical measures under 4 headings. The basic measures of level (G, Strength or relative level) and reverberance (EDT, early decay time and T60, reverberation time) along with the balance between clarity and reverberance (C80, early-to-late arriving sound ratio) are usually most important. This paper introduces the early relative sound level G80 and the late arriving relative sound level GL as useful diagnostic measures that can be derived from values of G and C80 (see Appendix). Finally, the lateral energy fraction, LF, (of the early arriving sound), the inter-aural cross correlation measures of the early and late arriving sound (IACCe and IACCl) are measures of spatial effects. The early and late arriving lateral sound levels (GEL and GLL) are also useful indicators of spatial effects that are not currently included in ISO3382.

Level Reverberance/ Clarity Diagnostic Spatial Effects G T60 G80 LF EDT GL IACCe C80 IACCl C50 GEL TS GLL Table1. Room acoustics measures discussed in this text. (See Appendix for definitions)

Many Frequencies: Measurements are usually made in the octave bands from 125 to 4k Hz. There are good reasons to extend this range to include the octave bands from 63 Hz to 8k Hz. We don’t yet have much knowledge concerning the preferred variation with frequency of the various acoustical measures but we do know that strong low frequency levels rather than long low frequency T60 values, influence the perceived strength of bass sounds in halls [1,2]. A full audio bandwidth is necessary for impulse response measurements if it is intended to listen to convolutions of speech and music with these impulse responses. Including the 63 Hz and 8k Hz octaves adds significantly to the requirements of the measurement system. It also suggests that testing laboratories should Consider measuring the acoustical properties of materials over this broader frequency range.

Measure ±10 cm ±30 cm G 0.4 dB 0.8 dB C80 0.6 dB 0.9 dB T60 0.04 s 0.06 s EDT 0.07 s 0.15 s LF 0.05 0.06

Table 2. Variations of measured values for small movements of the source or receiver [3].

Many positions: In a large hall one might typically measure at the 36 combination of 3 source positions and 12 receiver positions. This usually gives enough information for an understanding of spatial variations in a large hall. In a hall with more complex geometry, more positions would be preferred. Movements of the source or receiver position by as little as 10 or 30 cm can lead to measurable differences as shown in Table 2 [3]. The differences over 30 cm are similar to the just noticeable difference (JND) values for these quantities [4,5]. However, JND values are usually derived from careful A-B comparisons under laboratory conditions. Differences noticed by listeners in actual halls are probably much larger. These variations over 30 cm probably represent the accuracy with which conditions at a particular seat can be

characterized.

Purpose of Measurements and Data Reduction Techniques: The purpose of room acoustics measurements might be: a) to compare with design criteria, b) to better understand acoustical phenomena, or c) to diagnose the cause of acoustical problems. The approach to the data reduction will depend on the purpose of the measurements. The following examples illustrate various possible approaches.

EXAMPLE 1, SOUND LEVEL DETAILS IN A MULTI-PURPOSE HALL Measurements of sound levels in Southam Hall of the National Arts Centre, Ottawa are used to examine the question of how many measurement positions to use and the details of level variations in this hall. Figure 1 illustrates 1 kHz G value contours from measurements at 145 positions on one side the main floor of this hall, plus duplicate values reflected about the centre line to represent the other half of the main floor seating. While there are strong variations of these sound levels from front to rear of this hall, the pattern of variations is not that complex and there is little justification to make such detailed measurements.

B D F H K M O Q S U W Y 5 10 15 20 25 Length, (1 m/row) Wi dt h, m -1.0 0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 G(1k), dB

Figure 1. Contour fit to measured 1 kHz G values.

While Figure 1 gives a more global view of the characteristics, Figure 2 illustrates the details that can be found at an individual location. For seat O2 near the centre of the main floor seating, Figure 2 plots G values and their components versus frequency. While total G values do not vary greatly with frequency, the dominant component does. At lower frequencies the late arriving sound (GL) is dominant and the seat dip attenuation [6] has greatly reduced the early arriving sound (G80). At higher frequencies G80 values are the dominant component. If one goes into more detail and examines the lateral component of the early and late arriving sound (GEL and GLL respectively), we see that the lateral arriving energies are relatively small parts of the total early and late energy values. One would therefore expect a very low sense of spatial impression at this location.

125 250 500 1000 2000 4000 -25 -20 -15 -10 -5 0 5 Gx , d B Frequency, Hz G G80 GL GEL GLL

Figure 2. Measured levels and early and late components at a centre main floor seat. Examining how the variations of sound levels with frequency vary throughout the hall can give a more complete understanding of the acoustical characteristics. Figure 3 shows measured G values with early and late components at 3 different locations in the same hall. Figure 3(a) shows values at a seat very close to the stage where the direct and early reflection energy (G80) dominates at almost all frequencies. The results in Figure 2 showed that near the centre of the main floor seating, the balance of early and late arriving sound had changed considerably from the results in Figure 3(a). In the centre of the second balcony (Figure 3(b)) early arriving sound (G80) dominates at mid and high frequencies and the seat dip attenuation seems now to be greatest at 250 Hz. However, in the third balcony, the Figure 3(c) results show more similar early and late sound levels with strong late arriving sound (GL) at low frequencies. Clearly the character of the sound must vary considerably from area to area in this hall. 125 250 500 1000 2000 4000 -10 -5 0 5 10 G G80 GL Gx , d B Frequency, Hz (a) 125 250 500 1000 2000 4000 Frequency, Hz (b) 125 250 500 1000 2000 4000 (c) Frequency, Hz

Figure 3. Measured levels and early and late components, (a) seat B2 close to the stage, (b) seat D2 in the second balcony, and (c) seat D2 in the third balcony.

EXAMPLE 2, HALL AVERAGES AND WITHIN HALL VARIATIONS

Figure 4(a) plots hall average EDT and T60 values from the Northern Alberta Jubilee Auditorium (NAJA) with the range of EDT values obtained from measurements in 31 large halls [2] and a possible design goal for concert hall EDT values. Average EDT values are

smaller than T60 values as is commonly found, and the measured averages are less than the proposed design criterion values. However, this is very much a multi-purpose hall and it is easily argued that somewhat shorter decay times than the ‘ideal’ would be appropriate. The large range of EDT values shows that there are clearly many halls with much larger or smaller values. These results give no indication of the unusual characteristics of this hall that are seen when decay time values are plotted versus source-receiver distance as shown in Figure 4(b). While the 1 kHz T60 values are quite constant throughout the hall (as is almost always the case), EDT values vary quite dramatically with distance. Presumably, the sense of reverberance would be very different at the rear of the hall compared with at seats near the stage. These results, as well as those for early sound levels, can be explained as due to the shape of the ceiling sending all reflections to the rear of the hall as illustrated in Figure 5. In order to appreciate the significance of these within hall variations, we can compare the results in Figure 4(b) with those from other halls. These comparisons in Figure 6 show that variations of EDT in this hall (NAJA) are greater than in the other halls shown. In some halls there is very little variation of EDT values, sometimes because the hall is quite reverberant and diffuse (BOS), and other times because of different concentrations of early reflection energy to one area of the hall (SWP) [7].

125 250 500 1000 2000 4000 0 1 2 3 4 Ideal T60 EDT Dec a y t im e , s Frequency, Hz (a) 10 15 20 25 30 35 40 45 50 0.0 0.5 1.0 1.5 2.0 2.5 T60 EDT Distance, m (b)

Figure 4. Measured values from NAJA, (a) hall average T60 and EDT values, and the range of EDT values from 31 halls, (b) variation of 1 kHz EDT and T60 with source-receiver distance.

0 5 10 15 20 25 30 35 40 45 0.0 0.5 1.0 1.5 2.0 2.5 3.0 EDT , s Distance, m BOS SWP SLZ NAJA

Figure 6. Straight line fits to measured 1 kHz EDT values in 4 halls: BOS Boston Symphony Hall, SWP Salle Wilfrid Pelletier (Montreal), SLZ Neuesfestspeilhaus (Salzburg), and NAJA, Northern Alberta

Jubilee Auditorium (Edmonton).

EXAMPLE 3, COMPARISON WITH THERETICAL PREDICTIONS

It is often useful to compare measurements with theoretical predictions to determine if there are unusual features. Barron’s procedure [8] for predicting relative sound levels of the early, late and total sound components is particularly valuable because sound levels are so critical to subjective impressions of halls. In well-accepted halls, measured levels tend to agree well with the predictions of Barron’s theory.

0 4 8 12 G, d B (a) 0 4 8 12 G, d B (b) 5 10 15 20 25 30 35 40 45 -4 0 4 8 12 G, d B DIstance, m (C) 5 10 15 20 25 30 35 40 45-8 -4 0 4 8 G8 0, dB DIstance, m (d)

Figure 7. Comparison of measured 1 kHz values and predictions using Barron’s theory [8], (a) G values in BOS, (b) G values in SWP, (C) G values in NAJA, and (d) G80 values in NAJA.

For example, the variation of 1 kHz G values with distance in Boston Symphony Hall (BOS), in Figure 7(a), agree well with predictions. At seats close to the stage, levels are a little higher than predicted, probably due to strong reflections from the stage enclosure. At seats under the balcony (solid symbols) levels are a little lower than predicted. Figure 7(b) gives an example where measured levels closer to the stage are much above predictions, probably due to the shape of the orchestra shell. This results in large variations of levels from the front to the rear of this hall. Figure 7(c) shows a hall where measured values tend to fall systematically below predictions at most seats. This is again due to the ceiling shape of this hall tending to send early reflections to the rear of the hall (see Figure 5). In support of this explanation Figure 7(d) shows that early arriving sound levels (G80) are indeed higher than predicted at locations at the rear of this hall.

EXAMPLE 4, DIAGNOSING ACOUSTICAL PROBLEMS

The Orpheum Theatre in Vancouver was originally built as a theatre and later converted to a dedicated concert hall. Acoustical measurements identified several significant acoustical problems [9]. Again hall average values do not reveal problems and a simple plot of levels versus source-receiver distance (Figure 8(a)) shows a large amount of scatter. The causes of this scatter can be identified by separating the data into that from positions obtained at balcony seats and those from seats on the main floor of the hall including seats under the balcony. Figure 8(b) indicates that there are 2 separate trends for the two different groups of measurements. Sound levels tend to be lower at seats under the balcony but values measured at seats in the balcony tend to agree reasonably well with predictions using Barron’s theory.

5 10 15 20 25 30 35 40 -1 0 1 2 3 4 5 6 7 G (1k ), dB Distance, m 10 15 20 25 30 35 40 Distance, m

Figure 8. Measured 1 kHz G values versus source-receiver distance in the Orpheum Theatre and predicted levels using Barron’s theory (solid line in Figure (b)).

The situation can be more completely understood by examining how early and late sound levels each vary with distance. Figure 9(a) shows that most, but not all G80 values, agree quite well with the predictions of Barron’s theory. There are 5 positions in the balcony where early sound levels significantly exceed these predictions (i.e. by more than 1 dB). However, there is no systematic difference in early levels depending whether they are measured at seats in the balcony or under the balcony. On the other hand, Figure 9(b) shows that late arriving levels are systematically different depending whether measurements are made at locations in or under the

balcony. Measured values at locations under the balcony do not agree well with predictions using Barron’s theory but at other locations there is quite good agreement.

5 10 15 20 25 30 35 40 -4 -2 0 2 4 6 GL (1 k) , d B G 8 0( 1k ), d B Distance, m 10 15 20 25 30 35 40 -8 -6 -4 -2 0 2 Distance, m

Figure. 9. Measured values of (a) early levels (G80), and (b) late sound levels (GL) versus source- receiver distance in the Orpheum Theatre. Solid lines are predictions using Barron’s theory [8].

There are two different acoustical problems illustrated in these results. First, the very large balcony overhang, produces very obvious reductions in the late arriving sound at locations under the balcony. This is seen to a much lesser degree in Figure 7(a) for BOS. However, there is no systematic reduction of early arriving sound levels at seats under balconies. The other problem evident in these results, leads to increased early sound levels at particular locations in the balcony. These are due to various concave surfaces that focus sound to specific locations in this hall and cause significant variations in early levels. Figure 10(a) shows the initial part of the impulse response at the location with the most extreme focussing effect and another (Figure 10(b)) with no obvious focussing effect. The two strong reflections in Figure 10(a) are stronger than the direct sound and are perceived as a localization of the source in the ceiling.

0.0 0.1 0.2 0.3 -6000 -4000 -2000 0 2000 4000 6000 R e lat iv e pr es s u re TIme, s (a) 0.0 0.1 0.2 0.3 X Axis Title (B)

Figure 10 Impulse responses from balcony seats in the Orpheum Theatre, (a) with strong focusing, and (b) without obvious focussing.

CONCLUSIONS

Because ideal values of the various parameters defined in ISO3382 are not well established, it is often useful to compare measurements with values from well-regarded halls. Of course, just because a hall is well liked, does not ensure that measured values of all parameters are near to ideal values.

Comparing measured values with theoretical predictions can also be very helpful and can identify unusual acoustical characteristics in halls.

While hall average values can hide large within-hall variations of acoustical characteristics, averages of measured values over sub-areas and plots versus source-receiver distance can reveal important within-hall variations of acoustical conditions.

Examination of the values of early and late arriving relative sound levels (G80 and GL) is recommended as a very helpful diagnostic tool for obtaining a more complete understanding of the subjectively important aspects of concert hall sound fields.

There is an obvious need for new research to develop a better understanding of the subjective importance of variations with frequency of the various room acoustics parameters and to try to determine preferred spectral characteristics for each of them.

REFERENCES

[1] Bradley, J.S., Soulodre, G.A., and Norcross, S., Factors influencing the perception of bass, J. Acoust. Soc. Am., 101 (5) Pt. 2, p.3135, (1997).

[2] Bradley, J.S. The Sound field for listeners in concert halls and auditoria, Computational Architectural Acoustics, Editor J.J. Sendra WIT, Press, UK. (1999).

[3] Bradley J.S., Halliwell R.E., Accuracy and reproducibility of auditorium acoustics measures, Proceedings of British Institute of Acoustics, Spring '88 Meeting, Cambridge, U.K., 10, Part 1, 339-406, (1988).

[4] Cox, T.J., Davies, W.J., and Lam, Y.W., The sensitivity of listeners to early sound field changes in auditoria, Acustica, 79, pp. 27-41 (1993).

[5] Bradley, J.S., Reich, R., and Norcross, S.G., A Just Noticeable Difference in C50 for Speech, Applied

Acoustics, 58 (2) 99-108, (1999).

[6] Bradley J.S., Some further investigations of the seat dip effect, J. Acoust. Soc. Am., 90 (1) 324-333, (1991).

[7] Bradley J.S., A Comparison of three classical concert halls, J. Acoust. Soc. Am., 89, 1176-1192, (1991).

[8] Barron, M., and Lee, L.-J., Energy relations in concert auditoria, I, J. Acoust. Soc. Am. 84, 618-628, (1988).

[9] O’Keefe and Bradley, J.S., Acoustical renovations of the Orpheum, Vancouver: I. Measurements prior to renovations”, Canadian Acoustics, 28 (1) 21-33, (2000).

APPENDIX. DEFINITIONS OF SOME ROOM ACOUSTICS MEASURES This appendix defines the principal room acoustics measures discussed in the results of this paper, including some quantities not included in the ISO3382 standard.

The sound strength (or relative sound level) G is measured using a calibrated omni-directional sound source, and is calculated as follows,

dB , t d t t d t log 10 G 2 10 0 2 0 10 p p =              

∫

∫

where p(t) is the instantaneous pressure in the measured impulse response and p10(t) is the response to the same source at a distance of 10 m in a free field.

The strength of the direct and early arriving sound over the first 80 ms, G80, can be defined with an equation similar to equation A1 with the upper integration limit of the upper integral set to 0.08s. The sound strength of the late arriving sound, GL, consisting of sound energy arriving at the receiver more than 80 ms after the direct sound can again be defined with an equation similar to A1 but with the lower integration limit of the upper integral set to 0.08 s.

The relative level or strength of the early arriving lateral sound, GEL, and the late arriving lateral sound, GLL, can be calculated similar to G80 and GL but using the impulse response obtained from a figure-of-eight pattern microphone with the null pointed towards the sound source in the upper integral of an equation similar to A1.

The balance between early and late arriving sound energy can be measured using C80 (and several other related measures found in the ISO3382 standard). C80 is defined as follows,

dB , d d log 10 80 0 ) C _     

∫

∫

The lateral energy fraction is defined as follows,

(A4) t t p t t p = , L , , d d LF ( ) 2( ) 080 0 0 2 080 0 005 0

∫

∫

/

Here pL(t) is the instantaneous pressure response in the measured impulse response using a figure-of-eight pattern microphone with the null pointing toward the sound source.

The early and late relative sound levels G80 and GL can be calculated from measured G and C80 values as follows,

dB G G C C , 10 1 10 10 log 10 80 /10 10 / 80 10 / 80       ∗       + = (A5) dB GL G C C , 10 1 10 10 1 log 10 /10 10 / 80 10 / 80       ∗       + − = (A6)

Similarly the relative level or strength of the early arriving lateral sound can calculated from LF, C80 and G values as follows,

dB LF GEL G C C , 10 1 10 10 * log 10 /10 10 / 80 10 / 80       ∗       + = (A7)