ROOMS: REVERBERATION TIME - INTERRUPTED

NOISE PRECISION METHOD

0. INTRODUCTION

The purpose of this NORDTEST method is to promote uniformity

in the methods and conditions of the measurement of reverberation
times in reverberation rooms, so that values determined by
different measurement set-ups are in close agreement. The
reverberation time data determined by the method described,
may be used in situations where high precision is needed.

The recommendations in this NORDTEST method are more restric-

tive and stringent than the respective recommendations in IS0
354, Reference (I), without beeing contradictory. Accordingly,
this method can be used as a complement to IS0 354 as well as to
other standards in situations demanding a reverberation time
measurement of high precision.

1. SCOPE AND FIELD OF APPLICATION

This document specifies a precision laboratory NORDTEST method

of measuring reverberation times according to the interrupted
noise method. The document establishes directions for reference
measurements which could be used, for example, as a common basis
for the verification of other reverberation time measurement
methods.

published by key words classification

NORDTEST rooms UDC 543.62
Tekniikantie 12 reverberation time
FIN-02150 ESPOO interrupted noise method
FINLAND test method
www.nordtest.org
 - 2-

This document applies to computer-controlled equipment and to

measurement set-ups where the decay evaluations are made by a
computer. The computer may be an internal microcomputer in a
seif-contained instrument or a separate minicomputer combined
with the necessary instruments for generation and analysis.

2. REFERENCES

(1) IS0 354-1985 Measurement of sound absorption in a

reverberation room.

(2) NORDTEST method NT ACOU 012 (1979)

Reverberation room - Suspended diffusers: Absorption
coefficients.

(3) J.S.Bendat & A.G. Piersol RANDOM DATA: Analysis and

measurement procedures. Wiley-Interscience 1971.

(4) IS0 5725-1981(E), Precision of test methods - Determination

of repeatability and reproducibility by inter-laboratory
tests.

3. DEFINITIONS

For the purpose of this NORDTEST method, the following definitions

apply

3.1 Reverberation time: The time that would be required for the
sound pressure level to decrease by 60 dB after the sound source
has stopped.

The quantity is denoted by T and expressed in seconds.

Note - This definition is based on the assumption that, in the

ideal case, there is a linear relationship between the sound
pressure level and time and that the background noise level is
sufficiently low.

3.2 Repeatability r: The value below which the absolute

difference between two single test results obtained using the
same method on the identical test situation, under the same
conditions (same operator, same apparatus and a short interval
of times) may be expected to lie with a specified probability;
in the absence of other indications, the probability is 95 %.
 - 3 -

4. EQUIPMENT
4.1 Generation of sound field

The sound in the reverberation room shall be generated by one or

more loudspeakers whose radiation patterns are as non-directional
as possible. Measurements shall be made with a loudspeaker in at
least two successive positions (at least 3 m apart) or with an
equivalent multiple speaker arrangement, the speakers not
sounding simultaneously unless driven by separate, incoherent
and equally strong noise sources.

The test signals shall be band-limited noise having a continuous

frequency spectrum with a bandwidth of at least one-third octave.

The level of the steady exciting signal before decay shall be

sufficiently above the level of the background noise to permit
evaluation of the decay curves as specified in 5.2. The power of
the electrical noise signal fed to the loudspeakers shall remain
the same during all. the excitations.

The duration of the steady signal before switching off should not
be less than 1 s.

Notes
1 - If the ensemble averaging specified in 5.1 is extended to
include all the source and microphone positions, the loudspeaker
units in each source position shall be identical.

2 - If wide-band noise and a real-time analyser is used to make

simultaneous measurements for all frequency bands, the average
sound spectrum in the room should approximate pink noise with
differences in sound pressure level less than 6 dB between
adjacent one-third octaves.

4.2 Receiving equipment

The receiving equipment should consist of one or more microphones

which are as non-directional as possible, the necessary amplifiers,
filters and a measuring system for reverberation time.

The recordings should be made in at least 6 microphone positions

located at least λ /2 apart, where λ is the wavelength of sound
 - 4 -

for the centre frequency of the frequency band of interest.

Continuously moving microphones may be used. The length of the
microphone path should be at least 3 λ and the recordings should be
distributed uniformly or randomly along the microphone path.

Only one microphone should be used at a time. The microphones should

be at least 0,5 m from any absorber or diffuser and at least 2 m
from the sound source(s).

If more than one microphone is used and the ensemble averaging is

extended to include all the microphones, then it is necessary to
consider the sensitivity differences by calibrating each of the
various microphone channels.

The recording system should be equipment adequate for determining

the average slope of the decay curve of the corresponding reverberation
time.

The apparatus for recording the decay in sound pressure level may

d exponential averaging, with a continuous curve as output; or

b) exponential averaging, with successive discrete sample points
from the continuous average as output.

The averaging time of the exponential averaging device should be

less than T/20 and preferably smaller than T/25 but larger than T/50.

For apparatus in which the decay record is formed as a succession

of discrete points, the time interval between points on the record
shall be less than 1,5 times the averaging time of the device.

Note - The averaging time of an exponential averaging device is

equal to 8,69 divided by the decay rate, in decibels per second,
of the device.

One-third octave filters shall be included in the receiving equipment.

The discrimination characteristics of the filters should be in
accordance with IEC Publication 225.
 - 5 -

4.3 Centre frequencies

The measurements should be carried out at the following centre

frequencies (in hertz):

100 125 160 200 250 315

400 500 630 800 1 000 1 250
1 600 2 000 2 500 3 150 4 000 5 000

4.4 Number of measurements

The minimum number of measurements required for each frequency

band is:

a) 96 decays from 100 to 250 Hz (for example, 8 decays in each

of 12 sound source/microphone combinations);

b) 48 decays from 315 to 800 Hz (for example, 4 decays in each

of 12 sound source/microphone combinations);

c) 24 decays from 1000 to 5000 Hz (for example, 2 decays in

each of 12 sound source/microphone combinations).

4.5 Diffusers

If the actual test presupposes a diffuse sound field, the test

chamber should be provided with diffusers. A suitable configuration
of suspended diffusers can be found by using the procedure described
in the NORDTEST method NT ACOU 012, Reference (2). If a system
incorporating a rotating diffuser is used, then the need for
multiple loudspeaker and microphone positions may be reduced if
it can be shown that the precision of the method is maintained.

5 EVALUATION PROCEDURE

Since a computerized measurement procedure can only represent the

decay curve as a succession of discrete samples, this situation
is the only one that will be discussed below.
 - 6 -

5.1 Ensemble averaging

For each combination of microphone and loudspeaker positions, and

for each one-third octave band, an ensemble averaging procedure
should be used, involving the superposition of several repeated
excitations of the room to obtain a single decay curve from which
the reverberation time can be evaluated.

Such an ensemble averaging should be made by averaging the

instantaneous mean-square pressure values. Preferably, the
ensemble averaging should also be extended to include all the
source and microphone positions. Such a procedure makes it
possible to reveal if the energy dissipation of the test chamber
is non-exponential, producing a decay curve with some significant
curvature.

If an ensemble averaging is made in each combination of microphone

and loudspeaker positions, the reverberation time of the room is
given by the arithmetic mean of the total number of reverberation
time measurements.

No single decays may be excluded because of the application of some

curvature criterion.

Note - Curvature problems can be diminished by using rotating

diffusers, randomiy distributed low-frequency absorbers and/or
multiple sources.

5.2 Evaluation range

The recommended evaluation interval of the decay curve is

illustrated in Figure 1.
 - 7 -

Evaluation
--J
4 +
LLJ
l 0 0 0 interval
> o I
LIJ - IA t I
,t 2 A A
- .
0
3
0
0

Lu 0 <20dB
i
,” Li 0 >,20dB

0
0
,I n-l
-0
n
1

0 ‘t ‘i
TIME (s) -

Fig. 1. Recommended evaluation interval of decay curve.

The following conditions should be applied to determine the samples

i = 1, 2, . . . n, which comprise the evaluation interval:
first sample: tl 1 'I and tl < 'I + At
last sample: Ll - L 1 20 dB and Ll - Ln 1 < 20 dB
n
noise margin: Lnh L + 15 dB
noise

The moment when the noise is interrupted is t = 0. T denotes

the averaging time of the short time averaging analyzer and At
is the sampling interval. L noise is the combined background noise
level of the reverberation room and the recording equipment.

If the measurement resolution of the frequency analyzer is

decreasing towards lower levels then it is also necessary to
determine a sufficient margin at the bottom level of the analyzer.
This margin should be used in the same way as the background
noise margin.
 -8-

5.3 Algorithm

The reverberation time should be evaluated by using linear

regression analysis and the least-squares fit procedure. This
means that the decay curve is approximated by a straight line

ci = a + b t (1)
i'

where a and b are the intercept and slope estimates, respectively,

of the straight line. Thus, by definition the reverberation time
estimate is given by

jG -60/b (2)

According to the least-squares fit procedure the intercept and

slope estimates should be determined as follows

a = L - b t (3)

b = ( F ti Li -nfL)/(F ti2 - n t-2) (4)

i=l i=l
_ n n
where L = c Li/n and t = C ti/n
i=l i=l

Algorithms for further analysis of the decay curve are given in

the Annex.

6. PRECISION

The precision of the test procedure can be defined by its

repeatability, as described in IS0 5725, Reference (4).

For the time being, insufficient information on repeatability is

available to give an assessment of it in the context of this
NORDTEST method. For the purpose of checking the repeatability
in a laboratory, an estimation can be made using the method
described in IS0 354, Reference (1). Reliable figures on
repeatability can only be found by following the procedure for
an inter-laboratory test, as specified in IS0 5725.

Precision and repeatability can only be achieved if the room

conditions are stable. Requirements for temperature and relative
 - 9 -

humidity during measurements are given in IS0 354.

7. EXPRESSION OF RESULTS

The reverberation times should be calculated and expressed to at

least two decimal places.

8. TEST REPORT

This method can only be used as a complement to existing building

acoustics standards. In normal situations a reverberation time
measurement is only one of the steps in a full test. Therefore, it
is sufficient if the final test report includes a reference to this
NORDTEST method.

The test report shall include the following information,

if relevant:
a) Name and address of the testing laboratory
b) Identification number of the test report
c) Name and address of the organization or the person who ordered
the test
d) Purpose of the test
e) Method of sampling and other circumstances (date and person
responsible for the sampling)
f) Name and address of manufacturer or supplier of the tested
object
g) Name or other identification marks of the tested object
h) Description of the tested object
i) Date of supply of the tested object
j) Date of the test
k) Test method
1) Conditioning of the test specimens, environmental data during
the test (temperature, pressure, RH, etc)
m) Identification of the test equipment and instruments used
n) Any deviations from the test method
o) Test results (use SI units)
p) Inaccuracy or uncertainty of the test result
q) Date and signature
 ANNEX

The traditional regression theories make it possible to establish

a confidence interval for T based upon the estimates a, b and L.
Assuming a normal distribution of Li given ti, it can be shown that
a 95 % confidence interval is given by (Reference (3)).

n 2
T = -60/(b + t /cc t - n t2> 112) (1)
n-2,0.025 'Lit i=l i

where t is the 2,5 percentage point of the Student t

n-2,0.025
distribution and

SL/t =
c n
?I
(Li- Li)2 / (n-2)
3
This confidence statement reveals how well the measured decay
1/2
(2)

displays a linear relationship and how much scatter there is due

to extraneous random effects.

Curvature phenomenon can effectively be studied by plotting the

residuals - Li versus i. Another method is to derive a value of
L
i
the square of the multiple correlation coefficient R2,

i=l L
SS due to regression =
R2 = 3 (3)
SS about mean
c" (L - L)2
i
i=l

2
where SS denotes the sum of squares. R can have any value from
0 to 1. A decay curve which is perfectly straight corresponds
to R2 = 1. Preferably, a permillage deviation from perfect
linearity is studied by determining

c = 1000 (1 - R2) Oioo (4)

Finally, a percent curvature of the decay can also be defined as

C = 100 (Tb / Tt - 1) % (5)

where T b and Tt are the calculated reverberation times of the

bottom half and top half of the 20 dB-decay curve.
 ANNEX

Both < and C are sensitive not only to monotonic curvature but
also to random fluctuations. Therefore one can always recommend
an examination of residual plots.