United States
Environmental Protection
Agency
Office of
Noise Abatement and Control
Washington, DC 20460
EPA 550/9 79-103
November 1979
Noise
Annoyance, Loudness,
and Measurement of
Repetitive Type
Impulsive Noise
Sources
-------
EPA 550/9-79-103
ANNOYANCE, LOUDNESS,
AND MEASUREMENT OF
REPETITIVE TYPE
IMPULSIVE NOISE SOURCES
NOVEMBER 1979
Prepared by:
L. C. Sutherland
R. E. Burke
WYLE RESEARCH
El Segundo, California 90245
Prepared for:
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Noise Abatement and Control
Washington, D.C. 20460
This report has been approved for general availability. The con-
tents of this report reflect the views of the contractor, who is
responsible for the facts and the accuracy of the data presented
herein, and do not necessarily reflect the official views or
policy of the EPA. This report does not constitute a standard,
specification, or regulation.
-------
PREFACE
The United States Environmental Protection Agency (EPA) was
charged by Congress in the Noise Control Act of 1972, as amended by
the Quiet Communities Act of 1978, to conduct or finance research
to investigate "...the psychological and physiological effects of noise
on humans and the effects of noise on domestic animals, wildlife, and
property, and the determination of dose/response relationships suitable
for use in decision making..." (Section 14(b)(l)).
Pursuant to and as part of this mandate, EPA has undertaken investi-
gations to determine and quantify subjective reactions of individuals
and communities to different noise environments and sources of noise. A
specific series of studies has been initiated to determine the best
methods for evaluating subjective magnitude and aversiveness to noise on
the basis of spectral and temporal properties, and to ascertain the impor-
tance of and means for including nonacoustical factors in the evaluation
of general aversion to noise. The overall purpose of this line of research
is to derive a more solid basis for assessing the aversiveness of noise and
the benefits of noise control.
The aim of the investigation described in this report was to perform
a detailed analysis of data pertaining to potential annoyance responses
that may be attributed to repetitive type impulsive noise. Specifically, a
program was undertaken (1) to review and evaluate the literature on
human subjective response to repetitive impulsive noise, and (2) to assess the
-------
need for and relative order of magnitude of a subjective impulse
adjustment factor that would better define effective level in terms of
annoyance reactions.
The report provides much useful information on the annoyance and
loudness of repetitive impulsive noise. Moreover, it is expected that
the results of the investigation will form the basis of future experi-
mental psychoacoustic work to derive, if appropriate, more precise correc-
tions factors or noise prediction methods to effectively account for the
inherent annoyance associated with impulsive noise. EPA believes that
further research and evaluation of data on the subjective effects of noise
will foster the development of techniques to demonstrate additional
benefits of noise control beyond that exhibited by currently used pro-
cedures. Fulfillment of this objective awaits further study within this
series. The results published in this report, however, do provide an
important step toward a more complete understanding of the phenomena of
human subjective response to noise.
The conclusions reached in this report regarding moderate level impulsive
noise are the authors' and do not necessarily reflect the opinions of the
individuals listed above. Moreover, the U. S. Environmental Protection Agency
does not endorse the findings of this investigation for use as a "correction
factor" applicable to impulsive type noise, nor have similar correction
factors been used by the Agency in past or current noise impact analyses.
OFFICE OF THE SCIENTIFIC ASSISTANT
TO THE DEPUTY ASSISTANT ADMINISTRATOR
OFFICE OF NOISE ABATEMENT AND CONTROL
U. S. ENVIRONMENTAL PROTECTION AGENCY
n
-------
ABSTRACT
This study was undertaken to evaluate subjective and objective aspects of moderate
levels of noise from impulsive sources. The study excluded evaluation of hearing damage
risk or annoyance from building vibration by high level impulsive noise, which were
covered by recent recommendations of the National Research Council, Committee on
Hearing Bioacoustics and Biomechanics, Working Group 69. While the study included
original investigations into some of the objective aspects of impulsive noise, a detailed
review of the literature on the subjective aspects was emphasized. Based on this available
literature, the annoyance and loudness from a wide variety of repetitive impulse noises
were evaluated These results were applied to the evaluation of impulsive noise from
a number of specific noise sources. Based on the most pertinent literature, it is ten-
tatively concluded that a subjective impulse correction factor of +7 dB applied to the
A-weighted equivalent sound levels of these types of repetitive impulsive noise sources
would better define their effective level in terms of annoyance reactions. No additional
correction is identified at this time for crest level or repetition rate. Research on sub-
jective correction factors for helicopter blade slap is also reviewed and potential
reasons for the smaller subjective correction factors (i.e., 0 to 6 dB) for annoyance
response to this type of sound are discussed. It is recommended that refinements to this
subjective correction factor be based on the use of standard loudness calculation methods
(Stevens Mark VII or Zwicker) modified to include provision for a shorter time constant
to reflect subjective response to short duration impulsive sounds.
The study also included a brief experimental evaluation of the measurement of a
wide variety of simulated repetitive impulsive-type signals varying in duty cycle, repeti-
tion rate, pulse frequency, and ratio of peak impulse signal level to continuous background
noise level. When repetitive impulses are measured using maximum values of A-weighted
(slow) readings on an Impulse Sound Level Meter, no objective correction is necessary in
order to measure, with an accuracy of ±1 .5 dB, the equivalent sound level (L ) of the
wide variety of impulsive signals investigated.
-------
ACKNOWLEDGMENTS
A program to develop techniques for evaluating the noise from selected impulsive
noise sources was initiated in December 1975 for the Office of Noise Abatement and Control,
U.S. Environmental Protection Agency. Mr. Jeff Goldstein served as technical monitor.
The problems addressed in this study encompass areas in psychoacoustics and acousti-
cal measurement technology which have received intensive study by many investigators for
many years. The authors wish to thank the following individuals who, during the conduct of
this study, provided much helpful information from research which they or their colleagues
have conducted.
Dr. Robert Gales, Naval Undersea Center, San Diego, California
Dr. William J. Galloway, Bolt, Beranek and Newman, Canoga Park, California
Mr. B.W. Lawton, NASA Langley Research Center, Hampton, Virginia
Dr. O. Juhl Pedersen, Technical University, Denmark
Dr. D.W. Robinson, National Physical Laboratory, England
Dr. Bertram Scharf, Northeastern University, Boston, Massachusetts
Dr. Paul Schomer, U.S. Army Construction Engineering Research Laboratory,
Champaign, Illinois
Dr. Milton Whitcomb, National Academy of Sciences, CHABA, Washington, D.C.
Dr. Robert Young, Naval Undersea Center, San Diego, California.
The valuable contributions which were made to Appendix A by Dr. Mark Lee, presently
of the Jet Propulsion Laboratory, Pasadena, California, and to Appendix B by Mark Montroll,
are also appreciatively acknowledged.
-------
TABLE OF CONTENTS
Page
1.0 INTRODUCTION 1-1
2.0 SELECTION OF A BASELINE METRIC 2-1
2.1 Definition of Impulsive Noise 2-1
2.2 Baseline Noise Metric ' 2-10
3.0 SUBJECTIVE RESPONSES TO IMPULSIVE NOISE 3-1
3.1 Loudness or Noisiness of Impulsive Sounds 3-2
3.1.1 A Model for the Hearing Process .... 3-2
3.1.2 Experimental Data 3-4
3.1.3 Subjective Response to Impulsive Pure Tone
Sounds 3-8
3.1.4 Subjective Response to Bursts of Noise . . . 3-17
3.1.5 Loudness Versus Noisiness of Impulsive Noise . . 3-21
3.1.6 Subjective Response to Complex Impulsive
Sounds 3-27
3.2 Annoyance and Other Subjective Response to Impulsive
Noise 3-33
3.2.1 Annoyance Response to Impulsive Noise . . . 3-34
3.2.2 Helicopter Blade Slap Noise 3-39
3.2.3 Loudness Versus Annoyance of Impulsive Sounds . 3-47
3.2.4 Other Subjective Effects of Impulsive Noise . . 3-50
4.0 CONCLUSION: SUBJECTIVE CORRECTION FACTORS FOR
EVALUATION OF IMPULSIVE NOISE 4-1
4.1 Subjective Correction Factor A 4-1
4.1.1 Subjective Correction Factors Based on Loudness
Response Data for Tone and Noise Bursts . . . 4-1
4.1.2 Subjective Correction Factors Based on Measured
Loudness of Real Impulsive Noise Sources . . . 4-2
-------
TABLE OF CONTENTS (Continued)
4.1.3 Subjective Correction Factors Based on Annoyance
4.1.4 Summary of Methods for Computing the Subjective
Correction Factor A
s
APPENDIX A - OBJECTIVE MEASUREMENT OF IMPULSIVE NOISE
APPENDIX B - ISO ROUND ROBIN TESTS
APPENDIX C - FREQUENCY SPECTRA OF REPEATED TONE BURSTS
REFERENCES
Part A -
Part B -
PartC -
Part D -
PartE -
Part F -
Part G -
Annoyance of Impulsive or Fluctuating Sounds
Noisiness and Loudness of Impulsive or Fluctuating Sounds
Detection or Perception of Impulsive Sounds
Speech Interference
Sleep Interference
Hearing Damage
Measurement of Impulsive Sounds
Page
4-3
4-8
A-l
B-l
C-l
R-l
R-5
R-10
R-ll
R-12
R-12
R-14
VI
-------
LIST OF TABLES
Table Page
No. No.
1 Typical Physical Parameters of Four Real Sources of Impulsive Noise 2-9
2 Index of Experimental Studies on Loud ness/Noisiness of Impulsive or
Fluctuating Sounds Indicating Experimental Variables Investigated 3-6
3 Description of Naturally Occurring Impulsive Sounds as Comparison Signals 3-29
in Evaluation Experiment by Fidell and Pearsons
4 A Summary of Literature on Annoyance Responses to Impulsive Noise
(Excluding Studies for Helicopter Blade Slap) 3-35
5 Summary of Recent Studies of Helicopter Blade Slap Noise Including
Summary of Subjective Correction Factor for Impulsiveness 3-40
6 Comparison of Several Predicted Subjective Correction Factors for
Annoyance Applied to the Four Impulsive Noise Sources 4-7
7 Summary of Subjective Correction Factor (A ) Estimated from Existing
Methods or Data, dB s 4-8
VII
-------
LIST OF FIGURES
Figure Page
1 Examples of Time History Envelopes of Nonimpulsive and Impulsive
Sounds 2-2
2 Examples of Time Histories of the Instantaneous Pressure from
Impulsive Sources 2-4
3 Physical Parameters of a Typical Impulsive Sound 2-8
4 Conceptual Illustration of Auditory Process to Show Characteristic
Response Times (r) in Various Elements Which Govern the Dynamic
Response of the Ear to Transient Sounds. 3-3
5 Measured and Normalized Values for Change in Signal-to-Noise Ratio
of Single Tone Burst as a Function of Burst Duration 3-9
6 Range of Measured Loudness Level - Duration Tradeoff Reported from
Various Studies to Indicate Possible Range of Uncertainty in Predicted
Loudness of 20 ms Pulse 3-12
7 Subjective Correction A for Repeated 1000 Hz Tone Bursts 3-13
8 Subjective Correction A for Repeated Pure Tone Bursts, Pulse Duration
20 ms, Repetition Rate = 25/Second as a Function of (a) Reference
Intensity, L (Ref) at 1000 Hz and (b) Frequency for the 80 dB Reference
Level eq 3-14
9 Time History and Fourier Spectrum of a Typical Impulsive Signal 3-16
10 Subjective Correction Factor for Loudness or Noisiness Response to
Short Bursts of Noise Bands Relative to a Reference Noise 3-18
11 Difference Between the L of a Repetitive Noise Burst Superimposed
on a Steady Background Noise and the Level of a Continuous Noise
which Sounds Equally as Loud as a Function of the Ratio of the Burst
Level to the Background Level 3-19
12 Time-Amplitude Sequence Diagram of the Stimulus Presentation 3-22
13 Results of Experiment I 3-23
13a Comparison of Loudness and Noisiness versus Repetition Rate for a
Burst Time Fraction of 0.063 3-25
VIII
-------
LIST OF FIGURES (ConHnued)
Figure Page
14 Preliminary ValidaMon of Assessment Methods 3-26
15 Comparison of Loudness Calculation Methods for Triangular Transients 3-28
16 Comparison of Time-Integrated Measures of the Impulsive Noises with
the Same Measure of the Equally Noisy Reference Sound 3-30
17 Correlation of Judged Degree of Helicopter Blade Slap Versus Crest
Level 3-43
18 Illustration from Two Groups of Helicopter Blade Slap Data That Rank
Order of Annoyance Does Not Correlate with Judged Impulsiveness 3-43
19 Comparison of Noise Levels for Equal Annoyance Versus Equal Loudness 3-48
20 Typical Transmission Response of the Outer and Middle Ear 3-51
21 The 1968CHABA Damage-Risk Criterion for Impulsive Noise Exposure
and a Proposed Modification for a Nominal Exposure of 100 Impulses
Per Day at Normal Incidence 3-54
22 Comparison of Measured and Estimated Values of the Subjective
Correction Factor A as a Function of Crest Level 4-10
s
IX
-------
1.0 INTRODUCTION
Under the mandate of the Noise Control Act of 1972, the Environmental
Protection Agency is charged with taking steps to abate sources of noise potentially,
detrimental to the public health and welfare. Implicit in this is the need to establish
the means for evaluating and monitoring the noise from impulsive noise sources.
This report excludes consideration of human response to and measurement of
high level impulsive sounds such as sonic booms, weapons fire, or quarry blasts. The
latter topic has been rhe subject of recent recommendations to the Federal Govern-
ment by Working Group 69 of the National Research Council, Committee on Hearing,
Bioacoustics and Biomechanics (CHABA). With this limitation in mind, a research
study was carried out to develop an interim method for the evaluation of moderate
levels of impulsive noise below hearing damage risk levels. The method was to be
compatible with the existing methodology currently in use by the Environmental
Protection Agency (EPA) for evaluating community noise impact. The investigation
was divided into three basic elements:
1. Selection of a baseline metric for evaluating impulsive noise to
which subjective and objective correction factors* could be applied
as necessary.
2. Review and evaluation of the literature on subjective effects of
impulsive noise with emphasis on data relating to annoyance,
noisiness, or loudness of repetitive types of impulsive noise.
*Throughout this report, the term "subjective correction factor" is used as a convenient
label for the difference between the subjectively effective and objectively measured
value of loudness, noisiness or annoyance as defined in the text. It is not intended
to imply that the values cited for these "correction" factors can be used without
careful consideration of their validity and applicability for practical evaluation of
real impulsive sounds.
1-1
-------
3. Based on this review, the development- of a suitable method to account
for subjective (annoyance) effects of impulsive noise utilizing suitable
measurement methods and currently available instrumentation.
This report presents the results of this investigation in the following sequence:
• Section 2 discusses the selection of the baseline noise metric used
throughout the study.
• Section 3, the heart of the report, reviews the literature in detail on
loudness, noisiness, and annoyance responses to impulsive sounds. Other
subjective effects are also briefly covered.
• Section 4 summarizes the overall findings in terms of the differential
subjective response between impulsive and nonimpulsive sounds.
Three appendices are also included, covering:
• Appendix A - Objective factors involved in the measurement of
impulsive noise. This includes presentation of results of a laboratory
test of various noise metrics obtained from a precision impulsive sound
level meter when applied to a wide range of artificially-generated
impulsive sounds.
• Appendix B - Summary of the results of an international Round Robin
test on response to and measurement of impulsive sounds recently
conducted by the International Standards Organization.
• Appendix C - Frequency spectra of repeated time bursts. This
appendix briefly illustrates the spectral content of various ideal
repetitive tone bursts which roughly approximate some impulsive
sounds.
1-2
-------
2.0 SELECTION OF A BASELINE METRIC
2.1 Definition of Impulsive Noise
Sounds can be defined as impulsive when they exhibit some form of rapid and
substantial variation in the envelope of the time history of the instantaneous peak
pressures. This envelope can be visualized as a line connecting the instantaneous
peaks of a noise signal as measured on a high-speed oscillograph. Examples of
envelopes of impulsive and nonimpulsive sounds, illustrating this qualitative definition,
are shown in Figure 1. Figure la shows the envelope of peak pressures for fairly
steady sounds from a stationary noise source such as an electric motor running at con-
stant speed. Figure Ib shows a noise with a noticeable fluctuation of the envelope. This
may simply be called an unsteady or fluctuating noise such as from a stream of highly
variable traffic.
The first step in defining a baseline metric for the impulsive sounds considered
in this report was to classify all types of impulsive-like sounds into categories. As
illustrated in the figure, most types of impulsive sounds fit into two basic categories.
Figures Ic and Id show envelopes of the time history for sounds in these two categories
that are clearly impulsive — Figure Ic illustrates a single impulse such as from a quarry
blast and Figure Id shows a repetitive impulsive noise source such as from an unmuffled
rock drill or drop hammer.*
There are clearly other examples which fall somewhere in between the time
history characteristics shown here. For example, the envelope representing the time
history of an aircraft may look quite similar to that of the single impulsive sound
except that the time scale is stretched out to many seconds instead of hundredth* of
a second. However, in order to take advantage of any useful research that could be
related to impulsive noise, investigations on subjective reactions to all of the last
three examples iJlustrated in Figure 1 were grouped into three categories according to
the type of sound as follows:
*The latter is a wheeled vehicle equipped with a hydraulically operated drop hammer
and Is used for demolition of road surfaces.
2-1
-------
I I sec. |
a) Steady Sound
/\
j I sec
b) Unsteady, or Fluctuating Sound
(__!0ms
c) Single Impulsive Sound
*~~~J V^^-J^^r^^J \^^ffJ\^f^J \^
~*~~\ p~*~\ p**"^ t*~~*~\ r^^^i tf^~
I-1
.IOOms_J
d) Repetitive Impulsive Sound
Figure 1. Examples of Time History Envelopes of Nonimpulsive (see a, b)
and Impulsive (see c, d) Sounds
2-2
-------
I — Repetitive Impulsive Sounds
II — Single Impulsive Sounds
III — Unsteady Sounds
This review of impulsive noises is necessarily broad and potentially applicable to
a wide range of moderate to low level impulsive sounds. To illustrate the concepts pre-
sented in this report pertaining to loudness and annoyance of repetitive impulsive noises,
four particular sources were selected as typical of impulsive community noise. These are:
• Truck-Mounted Garbage Compactors
• Drop Hammers
• Two-Cycle Motorcycles
• Rock Drills
Clearly, some of these sources can generate impulsive noise levels which may
represent a hearing damage risk to the equipment operator or an immediately adjacent
bystander. However, hearing damage aspects of impulsive noise are not considered
in any detail in this review. Under certain operating conditions or with suitable noise
control features, these noise sources may not emit what would be called impulsive
noise according to our qualitative definition (i.e., rapid and substantial variation
in the envelope of the peak pressure time history). However, according to our three
categories above, all four of these sources, when generating impulsive sound, will
fall into Category I, i.e., sources of repetitive impulsive sounds.
Typical time histories of the instantaneous signals for each of the above sources
are illustrated in Figure 2.* For garbage compactors, ignoring the steady noise of the
power source used for its operation, the impulsive nature of compactor noise will
consist of random or irregular impacts of metal against metal so that the term "repetitive"
must, in this case, be interpreted as including such an aperiodic or random repetition .
For the other three sources, however, one can expect that under any given operating
condition, the repetition rate will be fairly constant so that the envelope will exhibit
a definite periodicity. It should be pointed out that repetition rates of concern in
this report will fall below the auditory range, that is, below about 20 Hz.
*The time histories shown in Figure 2 were obtained from a small sample within each source
category. They are not necessarily representative of all equipment thjat fall within those
categories.
-------
a) Commercial Garbage Truck with Compactor
Figure 2. Examples of Time Histories of the Instantaneous Pressure
from Impulsive Sources
2-4
-------
5 ms
b) Drop Hammer
Figure 2 (Continued)
2-5
-------
2.5 ms
c) Two-Stroke Motorcycle
Figure 2 (Continued)
2-6
-------
10 ms
d) Rock Drill
Figure 2 (Concluded)
2-7
-------
A typical train of impulsive sounds is illustrated in Figure 3. The five physical
parameters important for describing impulsive sound are defined for purposes of this
report as follows:
• Crest Level - The difference in sound pressure level between the peak
and rms level of the noise. For a background noise with a normal
(Gaussian) distribution of instantaneous pressure, the peak pressure
may be considered as the value at about three standard deviations
above the rms value. This peak, which ideally is exceeded only 1
percent of the time for Gaussian noise, will be about 10 dB higher
than the rms value. Thus, the crest level should normally exceed
about 10 dB before a noise is considered impulsive.
• Duration - The amount of time that the envelope of the instantaneous
pressure exceeds the rms value.
• Period (if repetitive) - The time duration between two successive
impulses in a train of impulses.
• Spectrum - The frequency distribution of acoustic energy in the impulse.
• Rise Time - The time required for the impulse to rise from the back-
ground noise to the peak.
Crest Level
Rise Time
D: Duration
P: Period
Figure 3. Physical Parameters of a Typical Impulsive Sound
2-8
-------
Representative values for these impulsive noise parameters for the two-stroke motor-
cycle, the drop hammer, rock drill, and truck-mounted garbage compactor are listed
in Table 1 .* For these sources of impulsive noise, the crest level lies between 13 and 30 dB,
the duration varies from several milliseconds to half a second, and the period varies from 10
milliseconds to 1-1/2 seconds. A frequency range of 200 Hz to 2 kHz covers most of the
acoustic energy of the impulsive noise. This table provides a general indication of the
magnitude of the parameters which define the general physical characteristics of the impul-
sive noise sources considered in this report. However, this range of parameters, in fact,
includes many other impulsive noises so that research into subjective response to all of these
can be applied, in part, to the evaluation of subjective response to the four particular
sources identified in Table 1 .
Table 1
Typical Physical Parameters of Four Real Sources of Impulsive Noise
Impulsive
Noise
Source
Two -Stroke
Motorcycle
Drop
Hammer
Rock Drill
Truck-Mounted
Garbage
Compactor
Crest
Level
dB
13
30
19
19
Pulse
Duration
ms
2-20
300
10
500
Repetition
Period
ms
30- 100
1500
50
5000
Peaks in
Frequency
Spectrum
kHz
0.30-2
0.25- 1
0.040 - 0.400
0.200- 1
Typical
Rise Times
ms
2
10
2
50
*The values listed in Table 1 were measured from a small sample within each source
category. Although there is no reason to suspect that the values listed are atypical,
the reader should apply caution in generalizing the conclusions of this study as
necessarily representative of all equipment that fall within each source category.
^Although selected as a repetitive impulsive noise source for purposes of this analysis,
recent information as presented in EPA Report No. 550/9-79-257, Regulatory Analysis
of the Noise Emission Regulation for Truck-Mounted Solid Waste Compactors, indicates
that this feature may not be necessarily characteristic of the majority of truck-mounted
solid waste compaction units.
2-9
-------
2.2 Baseline Noise Metric
Some sort of baseline noise metric is necessary for evaluating these various
impulsive sounds. This baseline metric should be: (1) reasonably unambiguous, (2)
measurable with precision laboratory equipment, (3) measurable with standard sound
level meters in the field with suitable correction factors, (4) compatible with the
day-night sound level (L , ) or the equivalent (energy average) sound level
(L ) metric, and (5) able to provide a foundation for application of subjective
Sq
impulsive noise corrections to allow comparison of the subjective response to impulsive
and nonimpulsive sounds. The baseline metrics applicable to the Category I impulsive
sounds could take one of the following alternate forms.
• Sound Exposure Level - The time-Integra ted measure of the A-weighted
sound level is identified by the symbol L<-.
• Equivalent Sound Level - The equivalent sound level is the energy-
average of the integrated A-weighted sound level over a specified
observation time Tand is identified by the symbol L .
r 7 eq
• Peak Sound Level - The maximum instantaneous A-weighted sound
pressure level during a given observation time is identified by the
symbol LApk.
• Peak Sound Pressure Level - The maximum instantaneous unweighted
(linear) sound pressure level during a given observation time is
identified by the symbol L , .
All of these metrics are essentially unambiguous quantities measurable in the
laboratory and potentially measurable by some of the advanced integrating sound level
meters. Measurement of the peak levels (L^ , or L ,) with sound level meters
equipped with a peak-hold position is straightforward, providing the rise-time of
2-10
-------
the signal is greater than 50 /^secs. This corresponds to an upper frequency limit of
20,000 Hz for significant energy in the spectrum of the impulsive sound.
Intentionally excluded from the candidate baseline metrics are the other
quantities measurable on a sound level meter. Those which will be considered later
for application to measurement of impulsive sounds include:
• Slow Sound Level - The exponential-averaged A-weighted sound level
measured with a nominal effective (squared pressure) time constant of
1 second, identified, for this report, by the symbol L
r\ J
• Sound Level or Fast Sound Level - The exponential-averaged
A-weighted sound level measured with a nominal effective time
constant of 125 ms, identified, for this report, by the symbol L
r\l
• Impulse Sound Level - The exponential-averaged sound level measured
with a nominal effective time constant of 35 ms, identified by the
symbol L. .
Other noise metrics could have been considered, such as measures
of statistical distribution, L , where x is the percent exceedence level, or noise
x
pollution level (L[N.p)which attempts to account for subjective reaction to fluctuation
of a noise, These were rejected as not being directly compatible with current EPA noise
metrics and are not readily measurable on standard sound level meters.
Returning to the four candidate baseline metrics, the last two measures of
peak level may be rejected at the outset as unsuitable because they fail to fit directly
into EPA's time integrated measures of noise, namely, day-night sound level L
dn
and equivalent level (L ). In order to make a final choice, it is necessary to consider
the general nature of the noise signatures that may be involved. For example, the
typical noise exposure of an individual at any one place to garbage compactor noise
might consist of several minutes of exposure to a relatively random series of impulses
generated by the clanking together of garbage materials as they are compacted, super-
imposed over the rising and falling hum of noise from the engine which drives the compactor.
2-11
-------
The duration of the exposure can only be roughly estimated and will vary widely from
one site to another and from one day to the next. The sound exposure level of such
a varying noise exposure would also vary accordingly, making it difficult to utilize
for realistic noise evaluation or certification unless one observation time were arbi-
trarily fixed. In this case, however, an equally useful measure would simply be the
equivalent (or energy average) sound level (Le ) during the measurement period.
In contrast, during a passby of a motorcycle, the only unambiguous energy-
related measure of the noise signature received by a nearby observer would be the
sound exposure level (L ). It would be possible to normalize the sound expo-
sure level by a standard duration of, say 10 seconds to provide what would
amount to the equivalent sound level over 10 seconds (i.e., L (10 sec)) with the
same energy as the actual event. On the other hand, if noise certification tests of
motorcycles were to be applied to stationary vehicles, the equivalent sound level (L )
during the observation period would be a logical baseline metric.
For the drop hammer or rock drill, a typical noise signature could consist
of a relatively long period of exposure, on the order of an hour or more with many
periods of more or less continuous exposure to the repetitive impulsive sound. In this
case, again, the equivalent sound level (L ) during the observation period appears
suitable as the baseline metric.
Thus, with the one exception of noise exposure to single events, which are
conveniently defined by the sound exposure level, it appears that the equivalent sound
level (L ) is the logical choice for a baseline metric for the impulsive sources con-
eq r
sidered in this study.
The A-weighting inherently incorporated in this metric is expected to provide
a more accurate or a more consistent correlation with human response to low level
impulsive sounds than would be provided by a nonweighted (linear) sound pressure level.
As will be discussed later, this observation is also consistent with the observed loudness
or noisiness of low level sonic boom sounds. These have been shown to correlate best
2-12
-------
with frequency-weighted measures (i.e., loudness in phons) of the sonic boom energy
spectrum which deemphasizes the low frequencies as does A-weighting.*
It remained only to define the observation time upon which the average sound
•ill be based. For the general c
observation time T will be defined as
level will be based. For the general case, the equivalent sound level (L ) over an
L =
eq
_T
10log10
1 / / 2 / 2 \
T / PA (t) / P dt
1 J \ A w / o /
L 0 ... -l
, dB (1)
where
P (t) — instantaneous A-weighted sound pressure at time t, Pa
f\
P = reference pressure (20 IJiPa), and
T = observation period, sec
For prediction of the day-night sound level (L ), the L for the impulsive
dn eq
sound is evaluated for the daytime (L,) - 0700-2200, and for nighttime (L , ) - 2200 to
d dn
0700 hours. The normal 10 dB penalty factor would be imposed on L for the baseline
dn
metric, but the possibility of increasing this for the potentially even greater annoyance
at night of impulsive sounds can be left as an option to be defined upon the basis of
examining the available information on sleep interference from impulsive sounds.
For application to defining the L of repeated single events, the same tech-
nique employed for specifying aircraft sound exposure will be used in the form
L = L. + 10 log N - 10 log [ T/t ] , dB (2)
6O O
*Note that for high level impulsive sounds, such as from quarry blasts or artillery,
C-weighted levels appear to predict community response quite well.21, 147
2-13
-------
where
L,- = sound exposure level of one event, dB re 20juPa • sec
N = number of events during the time T
T = observation period in seconds
t = reference time of 1 second
The observation time T to apply in the measurement of the equivalent sound
level will depend on the application, ranging from a minimum of 1 second (corre-
sponding to the duration of reference sounds often used in laboratory evaluation of
impulsive sounds), to 1 hour for an hourly equivalent sound level (L j(h)), to 15 hours
for the day sound level (L.) - the energy average during the hours 0700 to 2200.
In summary, then, the baseline metric used in this study for evaluation of*
impulsive noise will be the A-weighfed equivalent sound level (L ) measured over
a time to be specified as appropriate for each source. This provides a baseline noise
metric that is compatible with the existing methods developed by EPA for evaluation
of noise impact.* By providing adjustment factors (nominally identified herein as
correction factors) to the L to account for any subjective effects and measurement
errors for impulsive noise, it will be possible to properly include impulsive noises in
EPA's evaluation of environmental impact of impulsive noise sources. This metric is
also considered appropriate for application to each of the three categories of sounds
defined earlier: (a) Category I - Repetitive Impulsive Sounds, (b) Category II - Single
Impulsive Sounds, and (c) Category III - Unsteady or Fluctuating Nonimpulsive Sounds.
*U.S. Environmental Protection Agency, "Information on Levels of Environmental
Noise Requisite to Protect Public Health and Welfare with an Adequate Margin
of Safety." EPA Report No. 550/9-74-004, March 1974.
2-14
-------
3.0 SUBJECTIVE RESPONSES TO IMPULSIVE NOISE
Subjective responses of people to noise can be conveniently grouped into three
general (and overlapping) categories:
• Health-Critical Responses
- Hearing damage
- Long-term medical or psychological effects other than hearing
damage
• Activity or Behavioral-Influence Responses
- Speech interference
- Sleep interference
- Task interference
• Attitudinal or Judgment-Influence Responses
- Annoyance responses
- Loudness (or noisiness) judgments
The primary concern for subjective responses in this report is in the last category
(i.e., attitudinal or judgment responses), and therefore that category is the only category
that has been reviewed in depth. An extensive bibliography has been compiled, however,
on most of the above categories and is included in the Reference section at the end of
this report. For convenience, the bibliography is arranged chronologically within
each of five general subjects; Part A, Annoyance of Impulsive or Fluctuating Sounds;
Part B, Loudness or Noisiness of Impulsive or Fluctuating Sounds; Part C, Detection or
Perception of Impulsive Sounds; Part D, Speech Interference; Part E, Sleep Interference;
and Part F, Hearing Damage. An additional subdivision Part G, for the references on
measurement of impulsive sound,is also included in this bibliography. While all of the
sources are listed in the bibliography, for convenience, only the principal ones of con-
cern for this report are cited as references in the main body of the text.
3-1
-------
3.1 Loudness or Noisiness of Impulsive Sounds
As will be shown later in Section 3.2, a correction to L to account for the
eq
annoyance of impulsive sounds can range from approximately 5 to 15 dB, depending on
the correction method. Clearly, such a wide range of correction factors is of little
value so that a more precise method for selecting a subjective correction factor is
desired. The extensive literature on loudness or noisiness of impulsive sounds was
therefore reviewed emphasizing experimental results as a more reliable basis, at this
point, for assisting in the selection of a subjective correction factor. In addition,
these basic experimental results on response to transient sounds are expected to assist
in defining optimum ways to monitor impulsive noise. Following the review in this
section of the available experimental results on loudness and noisiness of impulsive
sounds, information related to the annoyance of such sounds and comparison of
annoyance and loudness or noisiness is considered in the next section. First, however,
it is helpful to consider a simplified model for the auditory process as a framework for
examining the data relative to impulsive noise response.
3.1.1 A Model for the Hearing Process
A simplified conceptual diagram of the auditory system is illustrated in Figure
4 to assist in defining the principal features significant in this study. As indicated in
the figure, characteristic response times for the "acoustic" parts of the auditory chain
(i.e., up to the point in the inner ear where spectrum analysis occurs) are much less
than the "RC" time-constant inside the last box where the overall detection, integration,
and recognition of sound signals is assumed to occur. Even considering the lowest
reported value for this time-constant, it is still more than two orders of magnitude
greater than for the earlier parts of the auditory chain which must be able to respond
to instantaneous pressure changes at rates up to 20,000 times per second (r = 50 /Llsec).
The "RC" time-constant, on the other hand, only limits the ability to track the
envelope of a sound. Thus, experimental studies on response of humans to transient
sounds have focused more attention on this part of the hearing process and have
utilized the RC smoothing filter concept illustrated as one of the ways to empirically
model the results. We will consider the implications of the model illustrated in
3-2
-------
CO
CO
Incidenf Sound Reflected
From Nearby Boundaries
Incident Sound
D i re c H y from
Source
Brain
Analysis of Signal
into (Critical) Bands
T= RC « 13 - 1000 ms
^— Neural Transmissji
Lines to Brain
Figure 4. Conceptual Illustration of Auditory Process to Show Characteristic Response
Times (r) in Various Elements Which Govern the Dynamic Response of the
Ear to Transient Sounds. Note: the Range for the Value of "RC" Time-
Constant Reflects the Extreme Range of Observed Values, (After Bruel,
Reference 146)
-------
Figure 4 again later, but first let us examine the experimental data on loudness and
noisiness.
3.1.2 Experimental Data
The independent and dependent variables involved in the noisiness of impulsive
sounds may be categorized as follows:
Independent Variables (The Stimulus)
- Signal Format
• Repetition, Single or Multiple Impulses
• Signal Spectrum Tone, Narrow Band Noise, Complex or Wide Band
Impulsive Noise (the complex impulse includes the type of real
impulsive sounds of concern in this report)
- Signal Characteristics Varied
• Pulse Duration
• Pulse Frequency
• Pulse Repetition Rate (for Repeated Impulse)
• Spectrum of Total Signal
• Rise and Decay Time of Pulse
• Phase of Signal Components
• Ratio of Pulse Signal Level to any Background Noise
• Duration of Total Exposure
* Method of Signal Presentation (including sound field characteristics
for loudspeaker presentation)
3-4
-------
Dependent Variables (The Response)
- Thresholds
• Absolute Detection (absence of noise)
• Masked Detection (in presence of noise)
• Flutter or Fluctuation Detection
- Magnitude
• Loudness
• Noisiness
Although noisiness and loudness are listed as separate dependent variables for
subjective response to impulsive sounds, it will be shown that they may be taken as
essentially identical. However, as shown later in Section 3.2.3, the annoyance response
to impulsive sounds may, in some cases, be significantly different from a loudness or
noisiness magnitude response.
Utilizing the above framework of independent and dependent variables on
loudness or noisiness of impulsive (or fluctuating) sounds, an index of the pertinent
available literature is presented in Table 2 which covers most of the major experimental
studies of subjective response to impulsive or fluctuating sounds. It will be convenient
to briefly review the pertinent findings of these experimental studies by three general
groups according to the type of stimulus.
• Pure Tones
• Bursts of Noise
• Complex Sounds (real or simulated impulsive noise including
helicopter blade slap)
3-5
-------
Table 2
Index of Experimental Studies on Loudness/Noisiness of Impulsive or Fluctuating Sounds
Indicating Experimental Variables Investigated
No
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
41
42
43
44
45
46
47
48
50
51
52
53
54
55
56
References
Author
Hughes
Garner &
Miller
Munson
Garner
Garner
Miller
Garner
Garner
Niese
Green +
Hamilton
Pollack
Plomp +
M-Fodor
Niese
Small +
Port +
Sheeley +
Carter
Zwicker
Zepler +
Garrett
Ekman +
Stevens +
Zwicker
Pearsons
Pearsons +
Dubrovskii +
Johnson +
Bauer +
Year
(1946)
(1947)
(1947)
(1947)
(1947)
(1948)
(1948)
(1949)
(1956)
(1957)
(1957)
(1958)
(1959)
(1960)
(1960)
(1962)
(1963)
(1964)
(1965)
(1965)
(1965)
(1965)
(1966)
(1966)
(1966)
(1967)
(1967)
(1967)
(1967)
(1967)
Sinqle Impulses ' /
Tones
D,F
D,F,L
D,F,L
D,L
D,F
D
D
D,F
D
D,L
D
D
Narrow Band
of Noise
D,L
0
Complex
D
SB,L
Wide Band
'D,L
D,L
D
D
Measured '
AT
MT
LL
AT,LL
LL
LL
MT
MT
MT
LL
LL
LL
LL
MT
LL
LL
LL
LL,A
LL
Multiple (Repeated) Impulses (')
Tones
D,F
D,F
D,F,R,L
0
Narrow Band
D
D
R
SB
D,R
AC, BBS
Wide Band
Noise
D
D
D,R
L
AM,R
Measured
MT,AT
AT,MT
LL
LL
LL
LL,MT
LL
LL
N
N
FT
LL
GO
-------
Table 2 (Concluded)
References
No.
57
58
59
60
61
62
63 64
65
66
67
70
71
72
73
74
75
76
77
78
79
80
Author
Herbert +
Shepherd +
Rothausen +
Johnson +
Reichardt +
Relchardt
Fidell +
Ollerhead
Shipton +
Thompson
Leverton
F'jchs
Carter
Stephens
Carter
Boone
Leverton
Pederson
Gustafson
Terhardt
Fuller +
Year
(1968)
(1968)
(1968)
(1969)
(1970)
(1970)
(1970)
(1971)
(1971)
(1971)
(1972)
(1972)
(1972)
(1973)
(1973)
(1973)
(1974)
(1974)
(1974)
(1974)
(1975)
Sinql e Impu ses
Tones
D
D
D
D
D.
D
Narrow Band
of Noise
D
Complex
SB
SB
D
D
D
D
D,F,R,L
Wide Band
Noise
Measured
LL,A
LL
LL
LL
N,A
LL
LL
LL
LL
LL
1 Multiple (Repeated) Impulses
Tones
R,RT
D
D
D
D
AM,R
Narrow Band
of Noise
D/R/F
D
Complex
R
D,R
D
D
HBS
F
R,RT
R,RT
HBS
D,F,R,L
D
Wide Band
Noise
R,RT
D,R
L,AC
Measured
FT
LL,A
LL
N,A
LL
LL
LL
LL,A
LL,A
LL
AT
LL
LL
'LL
FT
A
CJ
(1)
Independent variable identified by abbreviated code under column headings
which define type of signal: pure tone/ narrow band of noise, complex signal,
or wide band noise.
(2)
Dependent variable measured identified by abbreviated code.
Abbreviation Code
Independent Variables
D - Duration
F - Pulse Frequency
L - Signal Level
R - Repetition Rate
RT - Rise (and Decay) Tim*
(+: .tal)
SB - Sonic Boom Signal
HBS - Helicopter Blade Slap
AM - Amplitude Modulation
AC - Aircraft Sound
Dependent Variables
A - Annoyance
AT - Absolute Threshold
FT - Flutter Threshold
LL - Loudness Level
MT - Masked Threshold
N - Noisiness
-------
3.1.3 Subjective Response to Impulsive Pure Tone Sounds
Following pioneering work by Bekesy in 1929 on Hie effects of duration on
loudness of tones, Hughes, Garner and Miller, Munson, and Garner
laid the groundwork for subsequent studies on loudness of single or multiple tone bursts.
Typical results of this early work are represented by the data of Garner and Miller,
shown in Figure 5. Figure 5a shows the measured signal-to-noise ratio at detection
threshold for a single tone burst of varying duration presented in the presence of a
«
wide band masking noise. Figure 5b shows these same results normalized according to
a simple empirical model for the auditory detection process corresponding to the output
of a resistance-capacitgnce (RC) circuit. The latter is driven by a signal (E^) which
is assumed to represent the detected envelope of the tone burst. If we assume that the
tone is just detected when rhe peak output of the RC network reaches some fixed thresh-
' old detection level (E ), then it can be shown that for burst durations (T), much less
o
than the rime constant, T = RC (analogous to the ear's time constant), the required
signal level increases inversely as the pulse duration decreases, or
(Required Signal Level, E,) =- (r/T) (Detection Threshold, E ) (3)
1 o
Thus, the product of the signal magnitude E, and the pulse duration T is a constant,
as given by
E • T=" E r = constant (4)
I o
Since the product of the signal magnitude and pulse duration is a measure of the "energy"
in the signal, this relationship is simply another way to define the so-called "constant
energy" law normally invoked to explain why, for pure tone bursts with a short duration
relative to the ear's time-constant, the required signal level for detection increases
3 dB for every halving of the burst duration. This very same result was also obtained by
Munson when a tone burst was adjusted in level to equal the loudness of a fixed
duration reference tone longer than about 50 msec. However, as suggested by Munson27
and many others subsequently, a simple "RC" circuit model for the ear's response to
transient sounds has a limited application.
3-8
-------
35
€
-o~ 30
O
-C
s
f.
o
° 25
3.
a
"o
Z
f 20
a
c
0)
i/i
"" 15
10
i 1 1 — i — r 1 1 1 — i
a) Measured Data
O
— ilA/li — °
"el *
.! K-T^I o
I
O
o
A 0
_
8
A
O
a O O
6
A A
8
i iii i iii
1 — M •
400 Hz
670
1000 ~~
1900
ise Measured in Terms
of Spectrum Level —
—
_
0 -o\
Threshold
for
n^~ — "---A ConHnuouF
^^"""'-".^ Tone
-o) -
1 (( , ,
T, DuraHon, ms
30
25
20
10
r i i i i i i i
b) Normalized Results Based on "RC" Filter on Output of Detector
R Detection Level
Signal
iignal i 1 £ , . JL
Envelope] [_ tVl T
I I L
I I
0 T
T = RC
T
Frequency Time-Constant
Hz ms
Predicted with "RC Circuit" Model
Constant Energy Law \
_/ N
I I I
I I I
0.01
0.1 1.0
T/T/ DurationAime Constant
10
Figure 5. Measured (a) and Normalized (b) Values for Change in Signal-to-Nc»!s?
Ratio of Single Tone Burst as a Function of Burst Duration (Data from
Garner and Milled)
3-9
-------
The practical implication of this model for impulsive noise is that it could offer
a way to select an optimum procedure for measurement of impulsive sounds by dupli-
cating, electronically, the ear's internal time-constant. Such a rationale is the basis
for the 35 ms time-constant selected for impulse precision sound level meters (see
Appendix A). Unfortunately, there are several complications in this simplistic model which
are brought out by the experimental data.
Reichardt and Niese, employing a subject panel of 50 people, found that the
loudness matching of a tone burst of variable duration against a fixed duration reference
tone, usually of the order of 1 second long, was a very difficult experimental task for the
average subject when the two burst durations were substantially different and led to a
great deal of data scatter not indicated by the smaller subject panels (four to six) io most
other studies. By using reference tone durations near the middle of the range evaluated
for the test tone, they found much less scatter in the loudness balances. On the basis of
their refined technique, therefore, they measured a time-constant of 30 milliseconds.
These refinements also included a careful selection of the temporal spacing and
duration of the test and reference signals to avoid possible masking or memory errors in
comparing a test and reference tone or to avoid what they termed "roughness" which was
observed when a rhythmically repeated pattern was used for the test or reference signal,
A9
particularly at pulse repetition rates on the order of 3 to 50 Hz (Reichardt ). This
qualitative measure, "roughness," may be important in the evaluation of impulsive
noise.
Other factors which can cause variation in the observed trade-off between
signal level and duration are: (1) the "energy law" fails either when the signal duration
T is so short that a substantial portion of its frequency spectrum falls outside the critical
29
band centered on the pulse frequency (Garner ), or the signal duration is much longer
than the ear's time-constant, (2) the apparent time-constant increases as the signal
*J A "7^
level approaches the threshold of hearing (Garner and Miller, Boone ), and (3)
the time-constant apparently varies with frequency, as implied by the data shown in
3-10
-------
Figure 5. It has generally been accepted practice, however, to assume that the time-
constant does not vary with frequency.
The lack of agreement between investigators on the time-constant still con-
tinues. A recent study by Boone on loudness of repeated short tone bursts in noise,
79
using 20 subjects, produced a value for the time-constant of about 110 ms. Terhardt
has suggested an RC time-constant of 13 msec to fit his unique measurements of the
detection of periodic sinusoidal modulation (which he calls roughness) of pure tones.
In summary, there is substantial evidence to support values for the time-constant
ranging from 13 ms to over 200 ms (see Figure 6). Because there is no apparent way to
resolve this issue unequivocally for this report, the only practical choice appears
to be to work with the existing recommendations or practice for the choice of time-
constraints in impulse precision sound level meters.
This lack of agreement on the auditory time-constant is most unfortunate for it
implies the potential for conflicting evidence about a subjective correction factor for
impulsive sounds. This point is illustrated in Figure 6 which shows the potential range
of the time-constant based on the range of experimental data relating perceived loud-
ness of an impulsive sound versus its duration. Thus, for a given duration of an impulsive
sound, the potential increase in the signal level to achieve a loudness equal to that of
a reference (nonimpulsive) tone can be substantial. Clearly, any correction factor
for impulsive noise must be based as much as possible on experimental data for subjec-
tive response to real impulsive sounds.
3-11
-------
0) 00
— ' ~o
-o
c •»
3 0)
o c
:D
Q_
c
0)
25
20
15
10
- 5
TJ
O
00
Possible Range
of Loudness
for 20 ms Pulse
Range of
Measured
Level-Duration
Curves
I
0.1 2 512 5 10 2
Pulse Duration, ms
5 100 2 5 1000
Figure 6. Range of Measured Sound Level - Duration Tradeoff (where the Pulse Is Judged
to be Equally Loud to the Reference Tone) Reported from Various Studies to
Indicate Possible Range of Uncertainty in Predicted Loudness of 20 ms Pulse.
(Note that the Time Constant Specified by I EC Sound Level Meter Specifications
for "Fast" Would Tend to Fall Near the Middle of the Range of Measured Data —
Adapted from P. Bruel, Reference 146.)
Pulse Repetition Rate
Another major variable studied in loudness tests of repeated tone bursts is the
31
repetition rate. A very definitive study in this area was reported by Garner. From
these results on repetitive tone bursts, a subjective correction factor for real
impulsive noises can be inferred. His experimental procedure consisted of
presenting, through monaural earphones to six subjects, a continually repeated pattern
of a steady 1 second reference tone, 1/4 second silence, a 1 second cycle of repeated
tone bursts, a 1/4 second silence, 1 second reference tone and so on. The repetition
rate of the repeated tone burst group was varied from 5 to 100 pulses per second, the
pulse duration varied from 1 to 50 ms, the pulse frequency varied from 125 to 8000 Hz
and the intensity level of the pulse varied from 20 to 100 dB. The subject varied
3-12
-------
the intensity of the tone bursts until he obtained equal loudness to the reference tone. In
most cases, the energy in each tone burst group was less than that of the equally loud steady
reference tone. The subjective correction A for equal loudness for these tone bursts is simply
the positive difference between the sound exposure levels of the reference and test signals.
Since the tone burst group and the reference tone each lasts for 1 second, the difference in
sound exposure levels is also the difference in equivalent sound levels (L ). This subjective
eq
correction factor is shown in Figure 7 for 1000 Hz tone bursts and covers the repetition
rate and pulse duration range indicated. The intensity of the reference level was 80 dB.
The typical variation of A with reference level and frequency is shown in Figure 8.
16
OQ
12
-------
CO
-o
0
-2
I \ I
a) 1000 Hz
20
I
j_
40 60
L (Ref), dB
eq
I
80 100
10
0
100
b) L (Ref) = 80 dB
5 1000 2
Frequency, Hz
10,000
Figure 8. Subjective Correction A for Repeated Pure Tone Bursts,
Pulse Duration 20 ms, Repetition Rate = 25/Second as a
Function of a) Reference Intensity, L (Ref) at 1000 Hz
and b) Frequency for the 80 dB Reference Level (from
Garner31)
3-14
-------
As exhibited in Figures 7 and 8, the subjective correction factor behaves in a
complex fashion, even for simple tone bursts. These figures indicate the potential
difficulty of developing any simple, general method for predicting a subjective correction
factor for more complex impulsive noises which have different spectra, rates of attack
or decay, or amplitudes. Nevertheless, Garner was able to readily predict his experi-
mental results, such as those illustrated in Figures 7 and 8, on the basis of two basic
factors:
1. The spreading out of the frequency spectrum of repeated tone bursts.
As a result, the side band frequency components of the repetitive
tone burst can fall into critical bands outside the one centered on
the tone burst carrier frequency.
2. The shape of the loudness growth function. Due to this unique shape,
the loudness of the side band components in each of the several critical
bands involved in this broader spectrum of repeated tone bursts can add
up to a greater loudness than the sum of their energies because, at noise
levels well above threshold, the relative loudness of a sound changes much
more slowly than the relative intensity (i.e., 2-to-l change in loudness for
a 10-to-l (10 dB) change in intensity). That is, the loudness of sounds is
roughly proportional to the sum of the loudness in critical bands and the sum
of these loudness values in the side band components will not decrease as
rapidly at frequencies removed from the tone frequency as the physical
energies in these side band frequency components of repeated tone bursts.
There is really nothing new here, of course; it is simply the basic concept of loudness
93
summation of complex sounds which has been developed into a fine art by Stevens, ,
QO QA
Zwicker, and Niese. However, application of these well-developed concepts
for loudness of sounds has had only limited application to impulsive sounds.
It is important to recognize that the concept of "startle" is not involyed in a
prediction that a weaker impulsive sound can sound louder than a stronger steady-state
sound. This simply results from the accepted concepts for simulating the loudness
3-15
-------
percept-ion of sounds. It remains to be shown that there may indeed be an additional
effect that makes impulsive sounds more annoying than indicated by their loudness.
Frequency Spectra of Repetitive Tone Bursts
A brief consideration of the frequency spectra of repeated tone bursts is in order
here since this plays such a primary role in the concept just outlined. As Figure 9 shows,
repeated tone bursts produce a spectrum centered at the frequency of the tone with side
bands above and below this frequency.
t Repeated Tone Burst
/'
_-»-x--''fv 1
L.
«;"¥•-» -.*.
Figure 9. Time History and Fourier Spectrum of a Typical Impulsive Signal
In Appendix C, it is shown that the inverse of the duty cycle of the pulse (r/T)
provides a qualitative indication of the number (N) of side band harmonics within the
nominal "1/2 power" spectral bandwidth. The more the repetition rate increases', the louder
the pulse train will be, if t.he total energy stays constant. With a very long
duration, and a very slow repetition rate, all the energy of the signal is concentrated in
a narrow range of frequencies. If this range falls within a critical bandwidth, then the
loudness varies as the signal energy within this band. On the other hand, if this signal
spectrum bandwidth is much greater than the critical bandwidth, then loudness of the signal
will add approximately as the loudness of energy in each critical band but with emphasis
on the loudest band.
The broadening of the spectrum of tone bursts beyond the frequency of the tone
itself introduces an inherent complication in evaluating subjective response to inter-
mittent sounds. This complication is overcome, in a sense, by using a test signal -
broadband random noise, which already has a broad spectrum. Thus, the spectrum of
3-16
-------
repeated bursts of wide band random noise differs from the spectrum of the uninterrupted
noise only at frequencies, which are generally infrasonic, corresponding to the burst
repetition rate. Spectra of single bursts of broadband noise are not significantly differ-
ent from the spectrum of the steady noise itself.
3.1.4 Subjective Response to Bursts of Noise
28 30
Although Garner and Miller carried out initial studies on response to
oz
bursts of broadband noise, Pollack presented the first extensive study utilizing pulses
of wide band noise. For our purposes, his results may be summarized as showing that
the difference between the L of a continuous noninterrupted broadband reference noise
eq
and the L of equally loud pulses of the same noise increased from 0 to +10 dB as the
eq
duty cycle of the bursts decreased from 1 to 0. 1. For duty cycles below 0. 1, the
difference remained approximately constant at +10 dB. Thus, the subjective correction
factor A would be +10 dB for duty cycles less than 0.1.
41
Small, et al, used a more conventional procedure of balancing loudness of
repeated bursts of noise of various durations against interspersed 1/2 second bursts of a
constant level reference noise. They found that when the sensation level of the reference
noise burst was 60 dB (a typical listening level), the level of an equally loud variable
duration test burst was constant for durations down to 15 msec and then increased by 12.5
dB for each 10-to-l decrease in duration for shorter test bursts. For our purposes, this
is equivalent to the subjective correction A increasing linearly, at a rate of +3 dB
per halving of test burst duration, from a value of zero for 1/2 second noise bursts
to a maximum of 15.2 dB for a 15 ms noise burst, and then decreasing linearly at a rate
of -0.75 dB per halving of test burst duration for shorter bursts. This assumes that 1/2
second is the time base for computing the L of the variable duration noise burst.
eq
Garrett has repeated the tests of Pollack and Small, et al, with very similar
results as shown in Figure 10 where the measured values of A versus ratio of test signal
duration to reference signal duration is plotted. Similar data on loudness of a short
burst of 2 to 4 kHz noise from Bauer is also included along with data on relative
3-17
-------
noisiness of short bursts of noise bands from Fidell and Pearsons. The latter show little
agreement with the other data but this may be due to the unique experimental tech-
nique employed (free-field presentation, interaction with a computer for signal pre-
sentation), and the "noisiness response" instead of loudness. The values for A in this
case are actually differences in measured A-weighted noise levels of the reference and
test signals. The effect of A-weighting on the short noise burst levels is not clear.
20
CO
-a
cr
a>
*Equal Noisiness or Annoyanc
(Free-Field)
0.01 0.1 1.0
Duration of Test Signal/Duration of Reference Signal
Figure 10. Subjective Correction Factor for Loudness or Noisiness Response
to Short Bursts of Noise Bands Relative to a Reference Noise.
(Solid and Dashed Lines Identify Mean Lines Through
Experimental Data which Vary ~ ± 2 dB about Mean).
36
Finally, returning to Pollack, one particular set of his data provides a good
model for examining loudness of more realistic impulsive sounds. These data were
obtained on the loudness of partially interrupted noise. This consisted of a continuous
background noise with a superimposed periodic increase in noise by amounts varying from
3-18
-------
0 to 45 dB. Figure 11 shows the resulting data obtained under one condition of a repeHHon
rate of 1 pulse per second (pps) and a burst duration of 1 ms. The ordinate defines the
loudness level of the composite signal relative to the loudness level for continuous noise at
the same intensity as the noise peak. The dashed line on the figure shows the computed
L for this noise signal to illustrate, again, that the equally loud impulsive noise has an
L substantially less than the L of a continuous signal with the same: maximum level.
eq eq
The resulting subjective difference factor A approaches a maximum value of about 10 dB
S 36
for a ratio of noise burst to background noise greater than 30 dB. Other data by Pollack
47
and Garrett on partially interrupted noise gave similar results as shown in Figure 11.
CO
~D
Q)
DO
0)
_p
~0)
0
-5
-10
-15
-20
-25
-30
-35
i i I I I
$—i Pollack (Figure 1 of Ref. 36)
Pollack (Figure 4 of Ref. 36)
O Pollack (Figure 5 of Ref. 36)
x Garrett (Ref. 47)
Drawn to Fit Data
Equal Loudness Level
O
i 0-
A - 10.5dB
s
Computed L
eq
10 20 30 40 50 60 70
Burst to Background Ratio, dB
Figure 11. Difference Between the L of the Repetitive Noise Burst Superimposed on a
Steady Background Noise and the Level of a Continuous Noise which Sounds
Equally as Loud as a Function of the Ratio of the Burst Level to the Back-
ground Level (Burst Duration = 1 ms, Repetition Rate = 1 pps)
3-19
-------
In summary, with the exception of the results of Fidell and Pearsons, the
experimental data on loudness or noisiness of short bursts of random noise show con-
sistent trends similar to the tone burst data in terms of order of magnitude values for
the subjective difference factor. Limited results on the'ear's time-constant from these
studies are not inconsistent with the values observed from the pure tone tests. Additional
support for the time constant values discussed in Section 3.1.3 is provided by data by Dubrovski
and Tumarkina on subjective perception of the relative loudness of amplitude-modulated
noise. They hypothesize a time-constant for the ear of 10 ms to explain their data -
a value similar to the 13 ms cited earlier for tests on the modulation threshold of pure
tones.
The application of a "time-constant" model again appears convenient to explain
experimental results. However, this device may indeed be misleading based on the
30
unique results and resulting hypothesis posed by Miller in his study of the delay in
detectability of a low level noise signal following the interruption of a higher level
masking noise. Based on his results, Miller suggests that:
"... . the auditory system as a whole does not have a fixed rate of decay
of so many decibels per second independent of intensity. Thus the
auditory system cannot be said to have a "time constant" in the sense that
this term is generally used, and we have been careful to use the term
"critical duration" instead. This is not to say that the mechanism of the
ear has no time constant, however. As in all mechanical systems there
is a finite time required for the ossicular chain and the cochlear fluids
to begin and to stop their motions. The mechanical time constants of
this system, however, are far too small to account for the 65 msec periods
of perceptual growth and decay.
"It has often been convenient to liken the auditory system to an integrating
circuit The evidence seems to show that the ear is not so much an
integrating device as it is a delaying device According to our
hypothesis, the growth of the perception of noise is the integral of the
3-20
-------
distribution of transmission times of the various pathways from the cochlea
30
to the higher center, and not the integral of the sound intensity."
3.1.5 Loudness Versus Noisiness of Impulsive Noise
None of the preceding studies cited on noise bursts employed standard loud-
36
ness calculation procedures to predict their results. However, both Pollack and
4/
Garrett used different empirical approaches based on weighting their noise burst
signals by a function related to the observed level-duration trade-off. Garrett was
particularly successful in predicting the loudness of 48 complex transient signals
consisting of repeated decaying sinusoids. Other examples of this approach are covered
in Section 3.1.6 on response to complex impulsive sounds. However, before considering
more complex impulsive sounds, let us examine one final (and very recent) study on the
loudness and noisiness of noise bursts.
A new and unique approach to the prediction of human response to impulsive
82
noise is provided by the work of Izumi. In order to examine the possible subjective
difference between loudness and noisiness, Izumi conducted a set of two laboratory
controlled psychophysical experiments. In the first experiment a periodically inter-
mittent pink noise signal was used to determine if there was indeed any difference
between these two subjective parameters. Upon finding a significant difference, the
second experiment was conducted so that an effective assessment method could be
established.
For the first experiment, consisting of two phases, subjects were asked to com-
pare, using the paired comparison method, the test signal (intermittent pink noise) with
a standard signal (continuous pink noise at 70 dBA). The signals were presented in a
fade-in, fade-out sequence as shown in Figure 12. In order to overcome any possible
error due to sequence bias, the stimuli were presented both signal first and standard
first for an equal number of times. The whole procedure was repeated for six different
burst-time fractions (BTF); the BTF being defined as the signal-on time divided by the
on-plus-off time.
3-21
-------
ANNOUNCE
STANDARD
COMPARISON
13 K. 15 17
2526
30 (sec)
Figure 12 . Time-Amplitude Sequence Diagram of the Stimulus Presentation
(from Izumi °2 )
During Phase I, the subjects compared the pair of signals in terms of their
relative loudness, i.e., how much louder (or softer) than the continuous signal is the
intermittent signal? During Phase II the same signals were replayed, but this time the
subjects compared them in terms of their relative noisiness.
The results of both phases were tabulated and compared with each other (see
Figure 13). From these results, Izumi concluded that ".... as far as periodically
intermittent sounds are concerned, loudness judgments and noisiness judgments are
significantly and systematically different. Therefore, loudness and noisiness shall be
considered as different attributes. "
Once he determined that the two parameters are indeed different, Izumi set
up his second experiment in order to arrive at a model which would accurately predict
the noisiness of an intermittent signal.
In this experiment the subjects were presented with signals with 25 different
BTFs. They were asked after each trial to rate the relative noisiness of the signals
as in the first experiment.
3-22
-------
« 10
-D
_C
1
Jl
•fe
eg
-D 5
J
D)
1
4-
-2 o
£ °
i —
•
•
•
••
— ••
—
:
•
•
«_
•
*•
•
•
t
A
•
•
•
•
•
••
•
•
,-•-
•
V
_1_
•
LEFT
RIGHT!
•
t
*•- ¥
.-. T
•
COL. :'
' COL.:1
•
•
••
•
• •
•
•
n
•
LOUDNI
NOISIN
•
•
•
•
•
!
T
•
ESS"
ESS"
•
•
•
•
•
i
16/250 63/1000 630/1000
31/500 250/1000 950/1000
TIME PATTERN
Figure 13. Results of Experiment I. Okisa (Loudness) Data and Yakamashisa
(Noisiness) Data are Comparatively Plotted. Filled Circles
Represent Mean Relative Burst Levels Judged by Each Subject.
Averages and Standard Deviations are Shown by Central Lines
and Rectangles on Both Sides (from Izumi °^)
Model
From the results of these trials/ Izumi developed what he calls the "Perceived
Noisiness Model of Periodically Intermittent Sounds 75-A."
-1ST
= 6log10BTF+(10log10RR+10)(l -e
off,
, dB
(5)
3-23
-------
where
L_. = relative A-weighted noise level of burst in dB
RB
BTF = burst time fraction, i.e., on-time/on + off time
RR = repetition rate per second
T r = off time in seconds
off
In order to test this formula, he predicted the value of L _ for the '25 intermittent
noises used in Experiment II. The L 's were calculated using nine different methods: peak
Kb
burst levels in terms of Loudness Level, Stevens LL(S); Loudness Level, Zwicker LL(Z); Perceived
Of.
Noise Level, PNL; and A-weighted noise level; A-weighted equivalent sound level; Pollack's
method; Garrett's method; noise rating number (NRN) as specified by ISO ; and Model 75-A,
proposed by Izumi.
The predicted levels were then compared with the experimental data. The results
are shown in Figure 14.* From these results, Izumi's Model 75-A appears to be the best
predictor. The other methods always underestimate the perceived noisiness of the inter-
mittent sounds.
Startle Effect
The major reason, according to Izumi, for the difference between loudness and
noisiness is the startle effect created by the intermittence of the sound. The startle effect
is based on three physical parameters of the signal: repetition rate, rise time and the
burst-to-background ratio.
In these experiments the rise time and the burst-to-background ratio were held
constant and only the repetition rate was varied. The contribution of repetition rate to the
noisiness-loudness difference was quantified and this information, shown in Figure 13a, was
used in the development of Model 75-A. According to Izumi, work is still necessary if the
contribution of the startle effect is to be understood.
*Figure 14 is a corrected version of the form published in Reference 82 which was
kindly supplied by Dr. Izumi.
3-24
-------
It will be pointed out later that one study of subjective response to helicopter
23
blade slap as a function of rate of slap has also shown a trend of increasing apparent
noisiness with increasing repetition rate although the range of "pulse rate" explored was
well above that (i.e., 10 to 30 pps) explored by Izumi.
10
CD
C
£T
=)
00
UJ
LU
cc
I I I TT|
o-NOISINESS"
•"LOUDNESS"
I i i
I
0.5 1 248
REPETITION RATE, PULSES PER SECOND
Figure 13a. Comparison of Loudness and Noisiness versus Repetition Rate for
a Burst Time Fraction of 0.063 (from
3-25
-------
10
5
CO
t3
c
a
O
UJ
o
B -5
1
0
a
D -10
UJ
1-
CJ
O
LU
oc
o. -15
-20
—
—
i7>
m
1
CL
M
_I
_J
a
0)
CL
_i
"a
0)
a.
7
m
•D
O
a*
a.
cr
_j
_^
u
•*-•
<-•
a>
L.
l_
0
o
Z
cr
z
f
L
1
J
J
t»>
"al
•a
o
Figure 14. Preliminary Validation of Assessment Methods. Errors of Prediction
are Calculated for 25 Intermittent Noises in Experiment II. Mean
Errors are Shown by Central Lines and Standard Deviations by
Rectangles on Both Sides (from Izumi82).
In summary, although this is only one study/ Izumi shows quite well that
noisiness and loudness are not the same subjective quantities when dealing with inter-
mittent sounds, and that the startle effect of the intermittent sound is a prime cause
of this difference.
3-26
-------
3.1.6 Subjective Response to Complex Impulsive Sounds
Early work on subjective response to more complex impulsive sounds other than
tone or noise bursts involved measuring loudness of short triangular transients such as
44 72 74
repeated gun blasts. For example, Carter, Carter, and Carter and Dunlop
explored the loudness and threshold levels of this type of transient, which had a pulse
duration of 1 ms, for varying rise times (.05 to 0.5 ms) and repetition rates (1 to
256 pps). The effect of repetition rate was adequately covered by a simple energy rule
(+3 dB increase in intensity to maintain loudness for each halving of duration). For
the highest repetition rate, the ratio of on-time to off-time never exceeded 0.5 and was
typically much less. As with all the preceding impulsive noise studies cited so far
(except Fidel I and Pearsons), earphone presentation was used. The loudness judgments
were made by comparison of a 3 second reference (white noise) signal with two impulses
separated by 1 second from each other and from the reference noise. For most of the
loudness tests, the reference noise was fixed at 15 dB above threshold (sensation level
of 15 dB) for each subject.
The principal result from Carter's work is the evaluation of alternate means of
predicting loudness of triangular impulsive sounds. For each repetition rate and rise
time, the loudness of the reference noise and impulse, at the "equally loud" intensity
levels, was calculdted from the signal spectra. The spectra were computed from the pressure
time history for the impulsive sounds and measured directly for the reference noises.
The four calculation methods analyzed were:
-j . , 87, 89
• Zwicker
93
• Stevens, Mark VI
103
• Perceived Noise Level
• A-Weighting
The average difference between the calculated loudness (based on the com-
puted spectrum) of the impulsive sound, which was judged equally as loud as the refer-
ence sound, and the calculated loudness of the reference sound was measured for all the
3-27
-------
combinations of rise time and repetition rates (306 cases). The results are summarized
in Figure 15 for these four methods in terms of this difference as a function of pulse
repetition rate.
-no
dB
! 1
- $
I 1
I 1
•
&• *
6
i i
i i
+
fc x
1 !
1
ot-
*
|
1
f
u
X
|
'(0
fc
1
1
*
_ 1
i 1 I
» t
/W 4.
1 1 1
1 1
& x
1 I
1 1 1
(c)
CH — _i
X« g XD
1 1 1
1 i !
X * *
1 1 1
1
O
x
i
i
0
X
i
i
X
1
1
4
x
i
i
o
x
i
'(b)
i
o
x
i
i i i i i i i
i i i i i i i
i
0
x
'(d)
xo
1 4 16 64 256 " 1 4 16 64 256
Repetition Rate, pulses per second
Figure 15. Comparison of Loudness Calculation Methods for Triangular Transients.
Variation about the overall mean (within each loudness method) of the
mean difference (over subjects) between the calculated loudness (Li)
of the triangular 1 ms impulse and the calculated loudness (L ) of the
reference noise, subjectively judged to be equally loud.
Deviation from zero is a direct measure of the error in each loudness
calculation method: (a) Zwicker, phons; (b) Stevens Mark VI, phons;
(c) Perceived Noise Level; (d) A-Weighted Level. The symbols
denote varying rise time (», 0.5 ms; +, 0.25 ms; o, 0.1 ms; and
x, 0.05 ms). (From Carter72)
Surprisingly, the loudness computed on the basis of the A-weighted levels
exhibits the least deviation about an overall mean. The Zwicker method was next in
accuracy. There is reason to doubt the general applicability of these results, how-
ever, as shall be seen when these loudness calculation methods are applied to other
types of impulsive sounds.
63
Fidel I and Pearsons investigated the influence of phase of harmonic com-
ponents on the judged noisiness of five different simple transient sounds corresponding
to (l)an ideal N wave, (2) an N wave with 1 ms rise and decay times, (3) a triangular
waveform, (4) a square waveform, and (5) a doublet or positive and negative sharp
3-28
-------
impulse. Power spectra for each basic waveform were maintained essentially constant
while phase was adjusted by a computerized waveform generator. No significant
influence of phase on subjective loudness was detected.
They also evaluated the subjective loudness of 12 actual impulsive sounds and
eight artificial sounds presented, as were all their signals, over a high quality loud-
speaker system. The characteristics of these 12 sounds are listed on Table 3. The
difference between a time-integrated objective measure of the sounds and the same
measure for the reference sound is shown in Figure 16.
Table 3
Description of Naturally Occurring Impulsive Sounds Employed as />
Comparison Signals in Evaluation Experiment by Fidell and Pearsons
Impulse
1
2
3
4
5
6
7
8
9
10
11
12
Standard
Duration
(msec)
300
150
425
450
580
480
180
400
600
180
1200
900
1000
Identification
Automobile Door Slam
Paper Tearing
Hand Clap
Two Bottles Clinking
Together
Chain Collapsing on Itself
Nocturnal Animal Noise
Squeaky Release of Air
Through a Valve
Balloon Bursting
Balloon Bursting
Automobile Horn
Simulated Sonic Boom
Basketball Bounce in Highly
Reverberant Environment
White Noise, 1 Second
Approximate
Spectral Characteristics
Peaks at 0.5 kHz
Near flat spectrum to 10 kHz
Rises and falls about 0.8 kHz
Highly leptokurtic at 4 kHz
Near flat spectrum to 1 kHz, falls
slowly at higher frequencies
Complex spectrum peaked at 0. 125
and 2. 5 kHz
Peaks at 0.8, 1.6 and 5 kHz
Peaks at 0.2 kHz
Peaks at 0.2 kHz
Discrete frequency peaks concentrated
between 0.3 and 1 kHz
Predominantly low frequency, falling
steeply from 0. 125 kHz
Energy concentrated between 0.2 and
1.6kHz
Octave Band from 0.6 to 1.2 kHz
3-29
-------
-5
+5
DQ
"O
+10
GO
CJ
O
+15
+20
+25
1 1
n
nD
—
o n
o
- °D-A n
Oa
ooc0-)0 o n
° 8^ """'
O -° 12. f
n/~>
-^ \^ ^MHHMa^
0
-------
The ordinal's specifies Hie difference between the average sound level (com-
puted from a mean square average of the digitized time history of the signal) and
the same measure for the equally noisy impulsive sounds. For the A-weighted measure,
this difference is identical to our subjective correction factor A and was equal to 12.5
dB with a standard deviation of 3.5 dB. The standard deviations for the other measures
were slightly greater, thus indicating the A-weighted average sound level was slightly
more reliable as a predictor of noisiness of these impulsive sounds. Note that these
impulsive sounds vary substantially in their characteristics; some may not be very impul-
sive. However, they are all essentially single events and not repetitive. The average
value of A observed, in this case, has considerably more validity than the values
given up to now for the following reasons:
1. It was measured with a loudspeaker presentation thus insuring that
realistic head diffraction effects are included.
2. The objective measurement of the average sound level should be
very accurate — they were performed by digital analysis of a
recording of the actual sound reproduction.
3. The sounds cover a variety of actual impulsive noises to which
the subjects can relate.
4. The instructions to the subjects asked for a judged noisiness but
prompted an annoyance response as well (i.e., the test instructions
defined a noisy sound as annoying, unacceptable, objectionable,
and disturbing if heard in the home during the day and night).
Loudness measurements of decaying sinusoidal transients similar to those used
78
by Garrett were carried out by Gustafsson but at sound levels from 95 to 117 dB.
While the results tend to substantiate those given earlier, the high noise levels used
place these data outside the area of interest for this study.
3.1.6.1 ISO Round Robin Tests
The most complete set of data on loudness of impulsive noises is provided by the
final results of an international cooperative Round Robin test program organized
3-31
-------
under the auspices of the International Standards Organization, ISO/TC 43/SC-l,
Study Group B, "Loudness of Impulsive Sounds." The final report, prepared by Pedersen,
et al, represents results from 22 laboratories and "close to 400 subjects." Addi-
tional detailed supporting data were reported by Shipton, Evans, and Robinson, from
the National Physical Laboratory, on the specific results from their tests with the
ISO Round Robin data tapes. Detailed information on findings of the ISO Round Robin
Tests, drawn from these two sources, is presented in Appendix B.
Although the tests consisted of an evaluation of subjective and objective cor-
rection factors for the following three types of impulsive sounds, results for only the
subjective correction factors for the first group are considered here.
Group I Nine quasi-steady impulsive noises recorded from
actual sources such as a teletype, pneumatic hammer,
outboard motor.
Group II Five single impulse noises, such as from a gun or
mechanical ram.
Group III Six 1 kHz tone pulses of 5 to 160 ms duration.
The sounds were presented to the subjects via loudspeaker in repeated A-B sequences and
matched, in loudness, with reference signals presented at three sound levels (55, 75, and
95 dB re 20 MPa). The overall grand average subjective correction factor, A , for all
reporting laboratories, nearly 400 subjects, and for the nine repetitive noises in Group I,
is 12.5 dB. The standard deviation over the nine average values for each noise is 0.9
dB. This is a highly smoothed statistical result since the variation between subjects for
any one level and test sound can be 10 to 15 dB. However, it is estimated that the
final result is reliable within ±1.5 dB. No estimate could be made of subjective
correction factors for the five single impulse sounds since the equivalent noise levels
for these sounds were not available.
3-32
-------
3.1.6.2 Loudness of Sonic Booms
The evaluation of the loudness of sonic booms provides additional information
pertinent to the subjective response to impulsive sounds. Zepler and Harel success-
fully predicted the relative loudness of sonic boom sounds by applying a loudness
frequency weighting to the Fourier energy spectrum of the simulated N waves. Johnson
and Robinson ' have extended this type of approach to successfully correlate the
annoyance response from explosive blasts and sonic bands as well as conventional air-
craft sounds on the same loudness scale. They utilized the S.S. Stevens, Mark VI,
93
loudness calculation method with a modification to extend its low frequency range
to encompass the strong, very low frequency energy inherent in sonic booms. This low
64 67
frequency deficiency in the loudness calculation methods has been observed by others.
However, this may not be a significant problem for the type of impulsive sources of concern
in this report.
A key element in Johnson and Robinson's approach is the use of a specific 70 ms
integration time for measuring the signal spectrum. This was intended to duplicate the
ear's integration time. Note that this is twice the value of the time-constant
specified for the impulse precision sound level meter. This is obviously a critical point
that will require careful consideration in the selection of an optimum impulsive noise
monitoring technique.
Johnson and Robinson applied a loudness calculation scheme to the prediction
of annoyance for impulsive sources. Are these two forms of human response (loudness
and annoyance) really synonymous? The answer, based on available data is that they
are not necessarily the same. This point is fundamental to describing impulsive noise
and deserves the more careful review taken up in the next section.
3.2 Annoyance and Other Subjective Responses to Impulsive Noise
Review of the existing literature dealing with annoyance due to impulsive noise
yields a wide range of approaches and results. These results from available studies,
3-33
-------
excluding those on helicopter blade slap, are briefly summarized in Table 4. Annoyance
of helicopter blade slap is considered later. Annoyance due to aircraft sonic booms were of
primary concern in about half of the studies cited in Table 4. While most studies attempted
1 12
to measure annoyance, the terms "unpleasantness" and "unacceptability" were also
used. It was assumed that these terms represented a similar measure of subjective
response. The qualitative descriptor "annoyance" is not well defined but may be
assumed to represent an overall subjective reaction to an impulsive noise stimulus.
This reaction may very well integrate not only the loudness or noisiness sensation but also
the response to other non-acoustic factors such as startle/ emotional content or intrusive
noise level relative to the existing background ambient level.
3.2.1 Annoyance Response to Impulsive Noise
A division of the references on annoyance responses into the three categories
of impulsive noise studies defined earlier helped in selecting only those applicable to
this effort. Category I, which is the principal concern of this report, covers the
"repetitive impulses" produced by two-stroke motorcycles, rock drills, pavement
breakers, helicopter blade slap, and other repetitive impulsive noise sources.
Category II, "single impulse," includes sonic booms and artillery blasts. Category III,
"unsteady noise, " covers traffic and subsonic aircraft noise and is actually more
concerned with noise "events" rather than with "impulses." In one sense, however,
the first and last categories are similar, differing basically in the time scale of and
between "events" and in the crest factor or ratio of maximum peak pressure to rms
pressure.
In the studies cited, correction factors were developed to account for annoy-
ance on the basis of one or more features of the impulsive sounds: number or frequency
of impulses, amplitude, fluctuation (rate of change in amplitude), and duration. Some
investigators proposed correction factors which were applicable to impulsive noise in
general, independent of its characteristics. ' ' Thus, Eldred and ISO
R1996 propose a 5 dB correction should be added to any community noise which is
3-34
-------
Table 4
A Summary of Literature on Annoyance Responses to Impulsive Noise (Excluding Studies for Helicopter Blade Slap)
Noise
Type
1
Repetitive
Pulses
II
Single
Author
Plutchik'
g
Keigh ley
. , 10
Anderson,
Robinson
Eldred
ISO-R 19961 °
Fucta
2
Broadbent,
Robinson
Borsky
Kry ter
c L 1°
Sc homer
CHABA2'
WG *69
Date
1957
1970
1971
1971
1972
1964
1965
1970
1973
1975
No. of
Subjects
4
1902
24
-
100
79
3000
-
-
-
Noise Source
Test
Tone Burst
Live Office
Recorded
Road Drill
Community
Noises
Recorded
Handclap
Jet, PropA/C
Sonic Boom
Live
Sonic Boom
Sonic Boom
Artil lery,
Surface Blasts
Sonic Boom
Arttl lery,
Blasts,
Sonio Boom
Reference
None
None
None
None
Tone Burst
Jet
None
Subsonic Jet
None
-
Parameters
Varied
Frequency
Repetition
Rate, dB
-
Duration and
Number of
Bursts
Source, Level ,
Site
Handclap SPL
Tone Burst
Duration
Level
Source
Overpressure
(psf)
-
-
-
C o ns ta n t
Duration
-
Bkgd Level
Exposure
Duration
-
Handclap
Duration
Duration
-
-
-
-
Measurements
Subjective
Just Noticeable
Unpleasantness
Acceptabi lity
Adjective Pair
(11 Point
Comparison)
Community
Complaints
Annoyance
Annoyance
Ra t i ng
Annoyance
Unacceptabil i ty
Annoyance
Complaints
Annoyance
Objective
Attenuation
In dB
Peak Index (PI)
Average dBA
L
eq
eq
LA
LPN
Overpressure
Probabil ity
of psf
LPN
CNR
LCdn
Results
Base Scale
-
Average
LA
L
eq
L
eq
LA
PN
-
EPNL
EPNL
Lc
impulse ...
.- .. \o)
Correction
-
3.52(PI)t/2,
dB
4 a, dB(d>
5 dB
-
"(e)
-
|(ALpN,,dB
-------
Table 4 (Concluded)
Noise
Type
II!
U ns tea dy
Noise
Author
f O
Robinson
Parry, 1 £
Parry
Fuller,15
Robinson
Matschat, '4
Muller,
Zimmerman
Date
1969
1972
1973
1973
No. of
Subjects
-
-
24
352
Noise Source
Test
Traffic/
Aircraft
Aircraft
Traffic
Live
Aircraft
Reference
-
-
None
None
Para me ters
Varied
-
-
LA(max)
Level,
Duration, and
Frequency of
Events
Constant
Mea su re me n h
Subjective
Objective
Theory
Theory
Event
Duration,
Frequency
"
Annoyance
Tolerabiltty,
Activity
Disturbance,
Annoyance
LNP
L, L , NNI,
eq
LNP
Results
Base Scale
L
eq
PNL
L
eq
L
eq
Impulse
r~ w J
Correction
2.56 a, dB
0 (Duration)
2.56 a, dB
2 (g)
/O *
Response Scale
Annoyance
Acceptabi lity
Annoyance
General
Subjective
Reaction to
Noise
CO
CO
(a) Based on concepts by Rosenblith and Stevens as cited in Reference 11.
(b) Except as noted, equal to subjective correction factor A , in dB, to correct for subjective response to impulsive noise.
(c) PI, the Peak Index, is the sum of number of impulses which are 5, 10, 15 and 20 dB above L in 1 minute sample.
(d) C = Temporal Standard Deviation.
(e) Sonic boom, with L- = 116 dB indoors equally annoying as jet noise outdoors with !.„.. = 110 PNdB.
(f) ALp.. = Difference in L-KJ, in PNdB, between impulsive sound and background noise, respectively.
(g) Correction proportional to mean square rate of level fluctuation.
-------
deemed impulsive, while ISO R 1999 recommends a fixed 10 dB correction be applied
to assess the hearing damage risk of impulsive noise.
In most cases, as noted in Table 4, the various subjective correction factors
developed from the referenced studies are added to the measured L of an impulsive
eq
noise in order to obtain the effective L of a nonimpulsive noise that will produce
eq
the same annoyance. For one method, however, the correction is not to be added to
L but rather to the Effective Perceived Noise Level (EPNL). It might be assumed
that this correction factor proposed by Kryter could also be applied to the L scale.
21 ec'
Another correction method (developed by CHABA Working Group 69) involves the predic-
tion of annoyance that may, in part, be due to building vibration induced by impulsive sounds.
Since this response is potentially quite important for large amplitude impulsive sounds
and since it is not treated in any of the other studies, it is also included in the table.
However, as clearly pointed out by the CHABA Working Group, the concept, based
on the use of a C-weighted L without further correction, was designed to be appli-
eq
cable only to single high intensity impulsive sounds with a peak sound pressure level
above 100 to 110 dB - well above the peak sound level range of concern in this report.
In investigations of repetitive impulses, Category I, much of the emphasis has
been on how the level of the impulses fluctuates over time, or equivalently, what is the
probability density function of the impulsive noise. Two correction factors, based on
o
fluctuation, identified in Table 4, are the peak index (PI), proposed by Keighley
for rating acceptability of office noise, and a measure proportional to the temporal
standard deviation, a, proposed by Anderson and Robinson for application to
impulsive noises superimposed on steady random background noise. The peak index
(PI) measures the number of impulses in the sampling period at various peak levels
while the standard deviation measures only the rms variation about the mean - the rate
of impulses is not accounted for.
For investigations of single impulses, Category II, the number of impulses in
the sampling period and the level of impulses above the background level are the
3-37
-------
main considerations. A scale which incorporates both types of corrections into one
number may be needed for general application to Category II type impulsive sounds.
In investigations of unsteady noise (i.e., Category III noise), the primary
concern appears to be in the degree and rate of noise level fluctuation. Matschat, et
al, use a time-averaged measure of the rate of change of level while Robinson, et
al, ' ' use the average standard deviation of the level. These are not unrelated
since the former can be considered an approximate measure of the mean frequency of
the latter. The duration of fluctuating or repeated single noise events (specifically
12
aircraft noise) was found to be inconsequential by Parry and Parry contrary to the
conclusions of Kryter and others. Although these conflicting conclusions about the
effect of duration have never been fully resolved, the currently accepted practice is to
assume that an energy summation of noise events should be employed in rating their
noise impact. Thus, for Category III noises, this is equivalent to a rule that the effec-
tive noise level will increase directly as the duration (or more exactly as 10 log
(duration)). For multiple events of fixed effective duration t , the equivalent duration
e
correction is 10 log t + 10 log N, where N is the number of events.
e
To summarize so far,- the previous studies on impulsive correction factors for
annoyance of other than helicopter blade slap lead to several choices for the form
and magnitude of subjective correction factor. The form varies from a constant value to
a variable dependent upon, for example, the relative magnitude, rate of occurrence,
or rate of fluctuation of the impulsive sound.
The magnitude of the correction factor for annoyance will vary widely according
to these concepts covering a range of as much as 30 dB. A more definitive evaluation
of the magnitude of A for annoyance for the four impulsive sources of concern for this
report is developed in section (4). It is shown that, based on several of the
concepts summarized in Table 4, fhe value of A varies from 0 to 28 dB; the latter
8 s
value is based on the use of Keighley's peak index concept and is probably too high.
More reasonable values of A are shown to fall in the range of 5 to 13 dB.
3-38
-------
3.2.2 Helicopter Blade Slap Noise
Helicopter blade slap is a troublesome impulsive noise source which has
81
received a great deal of attention as to causes and effects. This attention has
focused, most recently, upon the practical problems associated with noise certification
of helicopters. It is, to a large extent, this more recent work which is briefly reviewed
here. Major aspects of 14 studies involving measurements of subjective response to
helicopter noise are summarized in Table 5.
o
The first study, by Pearsons, did not consider blade slqp per se but only
attempted to rate various momentary noise descriptors as to their accuracy for predicting
the relative noisiness of helicopter sounds. As indicated in the last column of Table 5,
the Perceived Noise Level metric appeared to be superior over others. Leverton
made, perhaps, the first attempt to quantify a blade slap correction factor for helicopters
and found that the A-weighted noise level from nonslapping helicopters had to be
increased 4 to 8 dBA above the A-weighted noise level from helicopters with blade slap to
achieve the same annoyance in a simulated living room listening situation. This would
imply an average subjective correction for annoyance from blade slap of +6 dB.
.17
Munch and King found a subjective correction factor to the sound exposure
level to predict annoyance of blade slap that increased linearly from +6 dB to +13 dB
as the crest level of the recorded helicopter noise signature increased from 14 to 21 dB.
The correction factor did not increase beyond 13 for higher crest levels.
18
Berry, Rennie and Fuller evaluated methods of measuring relative impulsive-
ness of blade slap and found the following typical crest levels for varying degrees of
impulsiveness.
Slightly Impulsive Crest Level = 5-10 dB
Moderately Impulsive Crest Level = 10-15 dB
Very Impulsive Crest Level — 20 dB
3-39
-------
Table 5
Summary of Recent Studies of Helicopter Blade Slap Noise
Including Summary of Subjective Correction Factor for Impulsiveness
Investigator
Pearsons
Leverton
Munch
and
King
Berry,
Rennie ,
Fuller
Man-
Acoustics
Law ton
.~
(Ref)
Jan .
1967
(3a)
Mar.
1972
(He)
1974
(17)
^
Oct.
1975
(18)
July
1976
(18a)
Dec.
1976
(19)
Subjects
21
7
7
20
12
24
40
Noise Source
Recorded
Comparison
w/Jet Noise
Recorded
Recorded
Recorded
Simulated
Simulated
Acoustical
Simulation
w/Continuous
and Impulse
Noise
Parameters
Base Scale
Reference Jet
Noise
LA
L
LPN
LA'LS
LpN, EPNL
LC'
L
A'
LPN
Parameters Varied
LC' LA'
L , L
N PN
Dur , Dur
u 10 20
LA
L
s
Crest Level
Time Constraints
(used in Integrals)
A (max)'
Impulsiveness Rate
Level
Impulsiveness
No. of Sine Waves
per Single Impulse ,
Frequency of Sine
Waves, Impulse
Repetition Fre-
quency, SPL
Continuous and
Crest Factor
Measurements
Subjective
Noisiness
Comparisons
Annoyance
Comparisons
Extent of
Blade Slap
Degree of
Impulsiveness
Annoyance
Annoyance
Annoyance
Objective
LC' LA'
L , L
N PN
10' 20
M a (b)
Ls
Crest Level
LPN
Impulsiveness
Ls
Corrected)
EPNL
LC'
LA'
LPN
Results
Correction
_
4-8 dB
(Subjective)
6-13 dB
(Varies with Crest
Level)
-
<0
1 to 4 dB
Approximately
2 dB
Human Response
Order of Appropriateness:
PN' N' A' C'
Duration and pure-tone corrections did
not improve prediction.
Impulsive helicopters, subjectively 4-8
dB more annoying than nonbanging
helicopters .
No reliable correction between
annoyance and impulsiveness.
Impulsiveness was not overriding factor
in judging annoyance. Less impulsive
signals frequently judged rougher,
more irregular and less predictive.
Pilot study showed negative correction
(Not statistically significant).
Correction increased with degree of
subjectively judged Impulsiveness.
Three base scales underestimate
annoyance 2 dB.
GO
-tv
O
-------
Table 5 (Concluded)
investigator
Patterson,
Mozo,
Schemer ,
Camp
Galloway^
Ga lanrer,
Popper,
Perera
Leverton,
Southwood,
Pike
Powell
F d'Ambra,
A . Damongeot
Klump,
Schmidt
Sfernfeld,
Doyle
Date
(Ref)
May
1977
(22)
Dec.
1977
(23)
Dec.
1977
(24)
1978
(24b)
July
1978
(24c)
1978
(24d)
1978
(24e)
June
1978
(24f)
No. of
Subjects
25
20
40
?
91
60
28
25
Helicopter
Noise Source
Real
Recorded
Sound
Simulation
Recorded
Comparison
with Recorded
Fixed Wing
Aircraft
Recorded
Sound
Simulation
Real Com-
parison with
Real Fixed
Wing
Aircraft
Recorded
Sound
Simulation
Recorded
Sound
Simulation
Simulated
Broadband &
Recorded
Helicopter
Impulses - By
Earphones
Parameters
Base Scale
LS,
EPNL,
V LC'
L
D2
EPNL
EPNL
EPNL
EPNL
LPN
LA
LA
Parameters Varied
Base Parameters
Components of the
Simulated Sound
EPNL, L
EPNL,
Crest Level
EPNL,
Crest Level
Mix Broadband
Noise with Real
Helicopter
Impulse Signals
L. , ,, Degree
of Blade Slap
LA , Relative
Measurements
Subjective
Annoyance
Relative
Annoyance
Annoyance
Annoyance
Noisiness in
Terms of
Unwanted
Objective-
ness , e tc .
Annoyance
Annoyance
Annoyance
Objective
Base
Parameters
EPNL per FAR
Part 36, EPNL
plus High
Sample Rote,
Crest Level
EPNL,
Crest Level
EPNL,
Crest Level
EPNL,
Crest Level
LA at High
Sample Rate
LPN' LA
p
LA'
Crest Level
LA'
L - L
C A
Crest Level
Results
Correction
OdB
4-5 dB
4-5 dB
0-6 dB
Depending on
Crest Level
None established
(-2 dB for EPNL,
+2 dB for L. , ,)
A(max)
A =2.8dB(e)
1.5-2. 5 dB
2.5-5 dB
Human Response
No correction for blade slap required.
No substantial penalty for helicopters
compared to fixed wing aircraft. SEL is
equivalent predictor to L and EPNL.
EPNL underestimates annoyance by 4
to 5 dB. Impulsiveness plus repeti-
tion rote adjustments give good predic-
tion of correction factor.
Helicopters were rated 4 to 5 dB more
annoying than commercial jet aircraft.
Crest level correlated better with EPNL
thanA-Wtd. SPL.
Corrections are needed for both main
rotor blade slop and tail rotor.
Neither ISO proposal nor A-weighted
crest level correction adequately pre-
dicted noisiness. More impulsive
helicopter was judged less noisy than
less impulsive helicopter.
The ISO method of correction (defined
in Ref. 24a) is considered best annoy-
ance descriptor.
Correction based on relative annoy-
ance of 7 sec Samples in lab setting.
Correction based on Wyle interpre-
tation of authors' method of adjustment
data .
Footnotes:
(a) No blade slap data.
(b) Between impulsive and nonimpulsive helicopters.
(c) A "Blade Slap Factor" (BSF) is defined to rate blade slap strength.
(d) Thii report attempts to close the gap between simulated and real helicopter noise.
(e) An equation is given for EPNL correction that is related to a digitized set of samples. A high sample rote is used. See text of reference.
-------
Their data relating crest level and judged relative degree of impulsiveness are com-
pared in Figure 17 with similar data from Leverton and Munch and King. A
consistent trend is apparent indicating crest level is a reasonably good predictor of
18
relative impulsiveness. However, Berry, etal, found quite a different story when
they attempted to predict relative annoyance of blade slap noise with the same objec-
tive measure. Figure 18 shows the relative rank order rating of judged annoyance of
two groups of helicopter noises with varying degrees of objectively and subjectively
observed "blade slap. " The experimenters found that there was not a reliable cor-
relation between them; in fact, the subjects seemed more responsive to the "roughness"
quality of the sound than to blade slap per se as a measure of its annoyance. As was
pointed out earlier, Reichardtand Niese observed a similar problem relative to
subjective response to repetitive impulsive sounds. Thus, according to this
limited set of data, crest level may not be a reliable predictor for rating annoyance
of impulsive noises.
Mabry, et al, measured the relative annoyance of simulated and recorded
real helicopter sounds with varying degrees of blade slap in a laboratory setting and found
that duration corrected noise level (using the Perceived Noise Level or A-weigh ted noise
metrics) correctly measured the annoyance response with little or no additional correction
required for blade slap. However, simulated helicopter sounds with subjectively judged
"light, " "moderate, " and "heavy" blade slap were about 1, 2, and 4 dB more
annoying, respectively, in terms of EPNL values, than a reference nonslapping
helicopter simulation.
19
Lawton, in an extensive laboratory investigation of continuous noises and
simulated helicopter sounds,found that Perceived Noise Level, A-weighted level and
overall sound level measures of simulated blade slap noise all underestimated the levels
that would produce the same annoyance as a steady sound by about 2 dB. Patterson
22
et al, using real helicopters, found that no correction factor was required to correct
for blade slap when helicopter sounds were measured in terms of time-integrated
A-weighted level (Sound Exposure Level, L ) or comparable metrics such as EPNL.
3-42
-------
Vary
Impulsive
Moderately
Impulsive
Slightly
Impulsive
A O
Severe
Mild
Marg inal
None
Peak
O Munch and King
76*
A Leverton
17
18
A A
Borry, Rennie and Fuller
•Peak Within 100-400 Hz Bond
10
20
30
Crest Level (Peak/rms Level), dB
Figure 17. Correlation of Judged Degree of Helicopter Blade Slap Versus
Crest Level
6 --5
a 5
o
c
o
>^
o
i 4
~E
6
I 2
1 -
O Hover (5 Sounds)
[> Flyover (6 Sounds)
-3
-2
O
>
>
o
>
SlighHy Moderately
Degree of Impulsiveness
Very
Figure 18. Illustration from Two Groups of Helicopter Blade Slap
Data That Rank Order of Annoyance Does Not
Correlate with Judged Impulsiveness (From Berry,
Rennie and Fuller)'8
3-43
-------
Galloway, using both recorded and real helicopter sounds, evaluated
tentative proposals by the British, French and the U.S.A. to the International Civil
Aviation Authority (ICAO)and the International Standardization Organization (ISO)
for blade slap penalty factors for helicopter noise certification. These are intended
to account for subjective response to blade slap and were proposed to correct measured
Effective Perceived Noise Levels of helicopter noise. The observed penalty factors for
equal annoyance for eight real helicopter sounds was 4 dB and 2 to 5 dB respectively
for two different types of simulated helicopter sounds. When each of the objective
correction methods proposed by ISO, which provided a measure of sjgnal impulsive-
23
ness only, were adjusted by Galloway according to the rate of blade slap impulses,
they predicted the observed subjective correction factors quite well.
Galanter, et al, found subjective correction factors of 4 to 5 dB to equate
annoyance of helicopters with conventional jet aircraft when both are measured in
terms of EPNL. In other words, the EPNL of the helicopter sound would have to be
about 4 to 5 below that of the CTOL aircraft for equal annoyance. In more recent
studies, Leverton, et al, have explored both blade slap correction factors and a
potential additional subjective correction factor to account for the pseudo-impulsive
nature of tail rotor noise. For the former, a correction factor varying linearly from
0 to 6 dB as "crest level" varies from II dB to 20 dB is recommended to explain results
of subjective tests for blade slap annoyance. (Crest level, in this case, is measured
by the difference between the peak level in the 250 Hz octave band and the A-weighted,
Slow level.)
24c
An extensive series of field tests by Powell using 90 subjects exposed indoors
and outdoors to two different real helicopters and a small fixed wing propeller aircraft
demonstrated that:
1. No significant improvement in noisiness predictability of EPNL was
provided by either an ISO-proposed correction factor or an A-weighted
crest level correction for impulsiveness.
2. For equal EPNL, the more impulsive helicopter was consistently judged
less noisy than was the less impulsive helicopter (i.e., A was negative).
3-44
-------
The latter anomalous result might be attributable to the fact that the subjects
were asked to rate relative noisiness instead of relative annoyance although their
instructions implied unwanted ness, objectionability, etc. as measures of noisiness.
Based on Powell's data, a blade slap penalty factor to be applied to EPNL would
actually be negative. (The actual value was about -2 dB; however, the penalty, or
subjective correction, factor was about +2 dB when applied to maximum A-weighted
levels.)
24d
In contrast, the study, by d'Ambra and Damongeot, carried out to validate
the latest ISO proposal for computing a blade slap correction factor, shows a small
but finite subjective correction factor for annoyance of 2.8 dB based on the average
result for 20 flights and 60 subjects.
24e
In the study by Klump and Schmidt, subjective responses to short recorded
(17 sec) samples of helicopter sounds, presented in a laboratory setting, were measured
and consistent evidence was found for an impulse correction factor, when applied to
A-weighted levels, of about +2 dB.
24f
For the last study considered, by Sternfeld and Doyle, a subjective cor-
rection factor could only be estimated due to the unique experimental (method of
adjustment) and data analysis techniques employed so the results are not included in the
following summary. However, the values of A estimated from their study do agree
very well with the average of the other studies.
The findings from these helicopter noise studies, which are pertinent to this
report, can be summarized as follows:
• The mean observed blade slap correction or penalty factor (assumed
^ roughly equivalent to rhe subjective correction factor A ) was 3.3 dB
±2.7 dB for the 11 studies which measured this quantity directly. How-
ever, three of these 11 studies found essentially a zero or negative
correction. The maximum correction for moderate blade slap (i.e.,
crest level of 10 to 15 dB) was about 6 dB. The maximum correction
for severe blade slap (i.e., crest level about 20 dB) was 13 dB,
comparable to the values measured for a variety of nonhelicopter sounds.
3-45
-------
The methods recently proposed to objectively compute g blade slap cor-
rection factor do not appear to agree consistently with the correction
factors measured subjectively to account for annoyance of blade slap.
Galloway shows that improved results are obtained if some modification
23
is made to account for variations in the frequency of the blade slap.
He shows results from one series of tests indicating A can change from
s
about 2 dB for a slap repetition rate of 10 Hz to 7 dB for a rate of 30 Hz.
This effect may explain part of the wide range in measured correction
factors. This dependency on repetition rates in this frequency range also
suggests that the "correction factor" may, in part, arise from inherent
errors in perceived noise level computations for signals with significant
energy below 50 Hz.
The proposed objective means for predicting a subjective correction factor
depend on some means of measuring the relative impulsiveness. The pro-
posed methods vary from a simple measurement of the crest level of the
A-weighted noise level ' to more complex procedures involving
sampling the detected signal (e.g., instantaneous A-weighted level)
at a high rate (~5000 Hz) and computing a measure of mean square
fluctuation level from these samples.
Finally, it is desirable to attempt some degree of resolution of the
differences in blade slap correction factors that evolved from the various
studies summarized in this section. Any attempt in this direction must
first recognize the substantial differences in experimental techniques
involved in the studies. Perhaps most important of all was the variation
in signal presentation. It varied from presentation to subjects in a labora-
tory setting of simulated or recorded real helicopter sounds lasting for only
a short period or for a complete flyby^to exposing subjects in the
3-46
-------
field to actual helicopter flyby noise. A review of the various results
seems to indicate that any "impulse" correction factor may be partially
masked or substantially reduced in real field tests where subjects were
exposed to the relatively long duration of the helicopter flyby. Thus,
larger duration corrections which are, in fact, characteristic of
helicopter noise, may serve to partially mask out the potentially added
annoyance of blade slap. Thus, results of those studies on subjective
response to helicopter blade slap probably cannot be used directly to
accurately define the magnitude of a correction factor for impulsive
noise alone.
So far, results have been presented on measured subjective correction factors
to account for either the relative annoyance or loudness of impulsive sounds. The
next section attempts to show how these potentially different responses may be related.
3.2.3 Loudness Versus Annoyance of Impulsive Sounds
The limited data dealing with comparison of annoyance versus loudness
responses to impulsive noises come from controlled laboratory tests. In an early Jabora-
83
tory study, Reese, Kryter, and Stevens found some evidence, shown in Figure 19,
that high frequencies, above 2000 Hz, were somewhat more annoying than indicated by
their loudness. However, the data are limited and exhibit considerable scatter.
88
Parnell, etal, found no such indication in their studies of response to bands of noise.
8A
Niese also found no distinction between loudness and annoyance response for a wide
variety of steady-state sounds but did find a difference between loudness and annoyance
when one-third octave bands of noise were presented as impulses. Shepherd and
58
Sutherland found that judged loudness and annoyance responses to simulated sonic
booms were the same for all cases except for the highest values of rise time investigated
(i.e., 10 ms). In this case, a slight decrease in annoyance was noted relative to the
loudness response. This can be interpreted to support Reese, Kryter and Stevens' data
3-47
-------
CO
T5
20
10 -
s $
Q) O
6 "§
n
ll -10
-20
-30
1000
Frequency in Hertz
0000
Figure 19. Comparison of Noise Levels for Equal Annoyance Versus Equal Loudness
(From Reese, Kryfer, and Srevens*^)
3-48
-------
indicating high frequencies are more annoying at the same loudness. An increase in
sonic boom rise time would tend to reduce high frequency content and hence reduce
annoyance more rapidly than loudness.
Rothauser, et al, investigated both annoyance and loudness judgments of
a panel to recorded typewriter sounds. They found that for keystroke rates less than
10 per second, a typewriter noise that was adjusted to be equally loud as a reference
wideband noise with a similar spectrum, had to be decreased in level about 2 dB to be
judged equally annoying. This would indicate a +2 dB correction to loudness criteria
for repetitive impulsive sounds like typewriters at repetition rates less than 10 per
second.
. 13
Fuchs, in a brief study of response to single handclap sounds, observed
that his subjects rated the claps about 5 to 6 dB more annoying than an equally loud
tone burst of comparable duration.
To summarize, laboratory data do not clearly support a significant difference
between loudness and annoyance of nonimpulsive sounds, but there appears to be a
consistent indication that there is a small positive difference between the annoyance
and loudness of many typical impulsive sounds. An annoyance correction of +3 dB
to a loudness-based subjective correction factor appears reasonable for repetitive
impulses with a rate less than 10 pps with zero correction at higher repetition rates.
Summary
So far, several possible approaches to the development of a subjective cor-
rection factor A to be added to L to account for annoyance effects have been
s eq
suggested:
• Computation from previously developed impulsive noise - annoyance
correction factors as outlined in this section.
3-49
-------
• Estimation from data on loudness of impulsive noise in terms of the
impulse signal parameters such as duly cycle and ratio of pulse
amplitude to background noise (see Figure 10).
• Application of the ISO Round Robin or Fidel I and Pearsons' data to
define A (see Section 3.1.6.1 and Appendix B).
• Application of existing loudness computation methods (i.e., Stevens,
Mark VI or VII, or Zwicker), possibly modified for an annoyance
(startle) effect of impulsive noise to compute A .
• Application of the new approach suggested by Izumi (see Section 3. 1.5).
To a large extent, the data for subjective correction factors for impulsive
noise are based on artifical listening situations in a laboratory and are thus subject
to certain limitations. First, subjects who rated "annoyance," loudness or noisiness
of impulsive sounds normally did so only while concentrating on the listening task
and were not burdened with other stimuli or tasks. Secondly, no objective (e.g.,
physiological) measures of the subjects' response were made. Nevertheless, sufficient
information appears to be available to provide the basis for a subjective correction
for evaluation of impulsive noise. Before developing this, however,- it is desirable
to briefly outline the other effects of impulsive noise which have not been discussed
and which could conceivably influence the selection of a subjective correction.
3.2.4 Other Subjective Effects of Impulsive Noise
3.2.4.1 Impulsive Noise and Models for the Hearing Process
Returning briefly to our conceptual model for hearing illustrated earlier in
Figure 4, there are other features to this model'related to audition of impulsive noise
which have not been mentioned. The significant effect of head diffraction on
i
modifying the pressure-time history on an incident sound field that reaches the ear
3-50
-------
95
has been clearly reviewed by Shaw. Related models for acoustic resonances, in the
99
external ear, analyzed by Teranishi and Shaw, identify the major resonances which
will further modify the pressure signature transmitted to the middle ear. The combined
transmission response of all of these elements, including the middle ear, add up to a
major factor which shapes the spectrum of the pressure signal processed in the inner ear.
As shown in Figure 20, this influence will be dominant in the high frequency range
(above 1000 Hz) where many impulsive noises tend to have their dominant spectral
content. Thus, subject-to-subject variation in these elements of the auditory process
will be more significant in considering measurement and evaluation of impulsive noise
than is the case for most other major noise sources which tend to have their energy
concentrated at low frequencies.
1000 2000
Frequency, Hz
5000 8000
70
n
_.— '
.'
'
\
,
/•' \
\
\
r
*
' 1 I 1
: •>-.' i ! .>•*
Middle Earj ' ~^-±.'' ' \
i : !
! I ; i
L--S
2OO Hz 500 1000 2000 5000 8000 .
Frequency, Hz
Figure 20. Typical Transmission Response of the Outer and Middle Ear
(Adopted from Bruel)
The simplistic elements for the detection and processing of auditory signals,
illustrated earlier in Figure 4, do not really represent the more advanced approaches
to this subject such as represented by the more detailed studies on auditory detection
, 84, 85, 90,91, 94 TL. . , ^ t. , , • lU \ t- c
theory. This work has potential bearing on the selection of an
optimum "time constant" model for application to optimum methods for measuring
impulsive noise. For example, the choice of the same "time constant" for both build-
up and decay of transient sounds is not necessarily well-founded by either theory or
observation (e.g., References 30, 56 and 91).
3-51
-------
3.2.4.2 Speech Interference From Impulsive Sounds
Reference materials for the interference of speech from impulsive sounds are
listed in References 96 to 106.
An empirical analysis of speech interference from intermittent sounds, presented
in the EPA "Criteria" document, indicated that for steady and intermittent sounds of
the same Energy Equivalent Level (L ) the speech interference of intermittent sound
could be greater than that for steady sound under certain conditions. A more detailed
TOO
analysis of speech interference of intermittent sounds using ANSI Standard methods
indicates that intermittent sounds should always exhibit substantially less speech inter-
ference than a non-intermittent sound with the same L . This is due, in part, to the
fact that the ANSI Standard includes a positive noise on-time correction to the articu-
lation index obtained from a steady-state masking noise. Thus, speech interference
effects do not appear to be the basis for any positive impulsive noise correction factor.
3.2.4.3 Sleep Interference From Impulsive Sounds
The effects of acoustic stimulation on sleep depend on several factors:
1 . The nature of the stimulus.
2. The stage of sleep.
3. Instructions to the subject and his psychophysiological and motivational
state.
4. Individual differences, e.g., sex, age, physical condition, and
psychopathology.
Due to the complex nature of the effects of noise on sleep, no attempt will be
made to elaborate on the sleep interference from impulsive sounds. However, pertinent
material on this subject can be found in References 107 to 111. It should only be
3-52
-------
mentioned that- the current use of a 1 0 dB penalty for assessing noise exposure at night '
may not be entirely adequate for evaluating nighttime exposure to impulsive noise due
to the potential for greater disturbance to sleep.
3.2.4.4 Hearing Loss Due to Impulsive Sounds
Both the energy principle and TTS~ (temporary threshold shift 2 minutes after
cessation of noise exposure) have been utilized to derive damage risk criteria for
, • • 121, 130 A ,. . , ,. r . .
impulsive noise exposure. Any discussion on the divergence of opinion on these
two methods is beyond the scope of this report. However, the topic is covered in
119
References 112 to 130. The CHABA damage risk criterion (1968) and its later
130
modified version are shown in Figure 21 to indicate the general magnitude of the
acceptable pressure level as a function of impulse duration for a normal incidence
condition at a normalized repetition rate. Therefore, the evaluation of hearing damage
due to impulsive sounds involves the measurement of the peak sound pressure level and
its time history. So far as is known, no attempt has been made to relate the type of
predictive information concerning hearing damage risk of impulsive sounds, contained
in Figure 21, to nontraumatic responses such as annoyance or loudness.
3-53
-------
a.
o
CN
CO
O
oo
165
160
155
150
145
140
135
130
125
i—i—i—i—r
CHABA(I968)
A-DURATION
MODIFIED
CHABA
LIMIT
A-Duration:Time Difference «-^ ,.
\ $
ND E
x>v
^
\
B-Duration: Time Difference AD (+EF when a reflection is
T present)
I I
1
.02 .05 .1 .2 .51 2 5 10 20 50 100 200
Duration, ms
500 1000
Figure 21. The 1968 CHABA Damage-Risk Criterion for Impulsive Noise Exposure
(Solid Lines) and a Proposed Modification (Dashed Lines) for a
Nominal Exposure of 100 Impulses Per Day at Normal Incidence.
Peak Sound Pressure Level is Expressed as a Function of A- or B-
Duration in the Range 25 Microseconds to 1 Second (Adopted
from Reference 130)
3-54
-------
4.0 CONCLUSION: SUBJECTIVE CORRECTION FACTORS FOR
EVALUATION OF IMPULSIVE NOISE
The approaches toward the development of subjective correction factors for
evaluation of impulsive noise are reviewed in this section and conclusions are drawn
concerning a method to account for the difference in subjective response between
impulsive and nonimpulsive sounds. The method is necessarily based on the type of
psychoacoustic response data available for impulsive sounds and does not necessarily
include other aspects of the subjective response due to factors such as startle effects or
emotional reaction to impulsive sounds.
4.1 Subjective Correction Factor A
The various approaches considered in Section 3 for the development of A
were based on: (1) computation from previously proposed annoyance correction factors;
(2) estimation from data on loudness response to impulsive noise; (3) application of the
ISO Round Robin ' ' , Fidell-Pearson or Izumi data; or (4) application of
some form of loudness computation method. These candidate approaches are compared
in this section.
4.1.1 Subjective Correction Factors Based on Loudness Response Data for Tone
and Noise Bursts
The laboratory data on tone and noise bursts can be used only for rough esti-
mates of the subjective correction factor due to the large difference between the test
signals employed and the real impulsive sounds of concern here. Nevertheless, based
on the limited information on the impulsive noise sources in Section 2, the following
rough estimate for A can be made. These estimates do not include any consideration
of a possible increase in annoyance response over loudness response.
4-1
-------
Basis for A
Noise Source Correction s'
Motorcycles Repeated Tone Bursts (Figure 7) 4-12
Drop Hammers Repeated Tone Bursts (Figure 7) 4-6
All Sources Repeated Noise Bursts (Figure 10) 4-14
All Sources Repeated Noise Bursts (Figure 11) 3-9
Mean of Range 4-10 dB
Attempting to estimate values of A from these data necessarily involves con-
siderable uncertainty and seems to indicate lower values than expected. However, it
31
should be recalled that in one case (Gamer ), the observed values of loudness for
repeated tone bursts were very well predicted by loudness calculations.
4.1.2 Subjective Correction Factors Based on Measured Loudness of Real
Impulsive Noise Sources
ISO Round Robin Data
The extensive ISO Round Robin data on A summarized in Appendix B lead to an
average value for A of 12.5 ±0.9 dB over all of the nine real impulsive sound tests.
Based on selecting values of A from the specific ISO sources that relate, approxi-
mately, to the sources considered in this study, the following estimates are obtained.
The values of A^ are rounded values from the ISO data in Table B-2 of Appendix B.
ISO Impulsive Source This Report Estimated Ag/ dB
Outboard Motor Motorcycle 13
Compressed Air Drill Rock Drill 14
Cement Mill Garbage Compactor! ., ,_
>••••••••« 11— I/
Mechanical Ram Drop Hammer ]
There is little justification for this attempt to pair-off the ISO and the four
specific sources identified in Section 2, since differences in noise signature may be
extensive. Thus, a single average number of 12.5 dB for A is considered a representative
result from the ISO data applicable to the sources considered for this report.
4-2
-------
Fidell-Pearsons Data
From Figure 16, the average A for the 12 impulsive noise sources listed in
Table 3 was 12.5 dB with a standard deviation of ±3.5 dB. Most of the 12 sources
studies differed substantially from those of concern here so that the direct applicability
of this value to the source in this report is questionable.
Izumi Data
As discussed in Section 3.1.5, Izumi has proposed a method for predicting an
effective burst level according to a noisiness response which seems to agree well with
subjective judgments (see Figure 14). Unfortunately, the parameters required by his
predictive model defined in Eq.(5) were not available with sufficient accuracy to
permit application of the model for this report. However, as noted in Figure 14, his
data do show that the average difference (A ) between the subjectively effective and
measured L for his 25 intermittent noises was 13.5 dB. This number may be compared
eq
to the value of 12.5 dB from Fidell and Pearsons.
4.1.3 Subjective Correction Factors Based on Annoyance
Several methods to directly account for the annoyance effect of impulsive
noise were outlined in Section 3.2. To determine both the applicability and the validity
of the various correction schemes proposed for repetitive impulses — shown earlier in
Table 4- corrections were calculated with some of these procedures for the four sources
of repetitive impulsive noise of concern for this report and for which data were available.
Three correction schemes, considered in Section 3.2, were applied to the real
impulsive noise sources: (1) Crest Factor (or Crest Level when expressed in decibels);
(2) Peak Index; and (3) Standard Deviation.
Crest Level Method
The first correction scheme, based on the Crest Level (C.L.), has been
previously proposed to predict a helicopter blade slap (subjective) correction factor
17 23
by Munch and King and Galloway , as discussed earlier in Section 3.2.2. For
Munch and King, A was given by:
4-3
-------
OdB for C.L. < 14 dB
From Munch and King, A = { C . L. - 8 dB for C.L. = 14 - 21 dB } (6)
13 dB for C.L. >21 dB
where C.L. = Crest Level = L. (peak) - L. (rms), dB
23
Galloway's results can be used to define two different predictive models
for A . The first is based on only the Crest Level (C.L.) for A-weighted levels. The
second is based on the addition of a pulse repetition rate (i;-J modifier. Both of these
"models" are based on the psychoacoustic tests conducted by Galloway and on his
regression analysis. Acknowledging the preliminary nature of these results as pointed
out by Galloway, they can be used to predict values of A as follows:
From Galloway
(Crest Level only) A = -4 + 0.54 (C. L.) , dB (7)
s
(Crest Level + Repetition A =-5.9 + 0.46 (C.L.) + 0.19 (i/) , dB (8)
Rate) S °
where C.L. = Crest Level = L. (peak) - L. (rms), dB
and v_ = pulse repetition rate, Hz.
Based on Galloway's rather limited data, which covered a range of 13.5 to 16 dB for
C.L. and about 11 to 25 Hz for v~, his first expression could be replaced, for all
practical purposes, by a simple linear equation, A -C.L. - 11, dB, similar to that of
Munch and King. Galloway also points out that the additive correction term for
repetition rate is expected to reach a maximum value at 30 to 40 Hz and then decrease
at higher repetition rates.
The Crest Level for the impulsive sources considered in this report was measured
in the following manner. The tape-recorded test noise was fed into an impulse precision
sound level meter (B&K 2204/5) and the highest value of the A-weighted SLOW response
4-4
-------
was read to define the maximum rms level (L (rms)) of the impulsive sound. The same
/*\
signal was monitored on an oscilloscope and the maximum peak level L. (peak)
r\
determined. The repetition rate was estimated from oscillographic records of the
four noise sources (see Table 1, Section 2.1). The preceding expressions were then
used to compute values of A . No attempt was made to apply the other impulsive noise
S r\ A
correction method proposed by ISO for helicopter noise, which requires that the A-
weighted noise signal be sampled at a rate of 5000 samples per second.
Peak Index Method
0
The Peak Index correction method, proposed by Keighley for office machine
noise, was applied to each of the sample impulsive noise sources, with the exception of
the motorcycle. The value of the Peak Index (PI) was derived by examining the time
histories of the impulsive sounds on a graphic level recorder set to a writing speed of
125 dB/second. The number of peaks in 1 minute, which were at least 5, 10, 15,
and 20 dB above the average graphic level reading, were tabulated and summed. The
square root of this number, when multiplied by the constant 3.52, gives the value of
A for this scheme. Motorcycle noise was not evaluated with this method because the
time variation of the noise level was such that true peaks were not registered by the
graphic level recorder at the pen-speed setting used.
Standard Deviation Methods
The Standard Deviation correction, proposed by Anderson and Robinson for
general impulsive noises, was also obtained using the graphic recorder, and again
motorcycle noise was excluded because of its rapid time variation. Using a sampling
period less than the duration of typical impulses for the other sources, the average
levels for up to 100 successive periods were manually compiled from the graphic level
recordings and tabulated in a histogram. From the histogram, the Standard Deviation
of the A-weighted level was then calculated and multiplied by 4 to give the value of
A as prescribed in Reference 10.
4-5
-------
ISO R 1996 Method
The ISO R 19961 la correction of 5 dB is considered as a fourth method to be
considered for predicting A .
Results
Table 6 provides a comparison of the results of applying the preceding schemes
for predicting A for annoyance. For each impulsive noise source, the various values
of A allow comparison between the various methods, even though their absolute
validity remains dependent on direct psychophysical experiments involving the noises
themselves.
It has already been pointed out in Section 3.2.2, that some of the studies on
helicopter blade slap demonstrated that crest level was a reliable predictor of subjec-
tively judged impulsiveness but an unreliable predictor of annoyance. '
Hence, values of A based on this parameter alone may not be reliable. However,
when repetition rate is included, Galloway's data show a substantial improvement in
the ability to predict a value of A in agreement with the observed value. His cor-
s
relation coefficient increased from 0.42 (A predicted by crest level only) to 0.88
when the repetition rate correction was added, thus indicating the potential significance
of this parameter for subjective response to impulsive sounds.
The fact that the Peak Index correction scheme necessitates a 1 minute
sample may make it inapplicable for many passby or intermittent impulsive sounds. A
similar correction based on a 10 to 20 second sample may be more practical in such
cases. However, in the absence of any other supporting data, this method for predicting
A is not considered further in this report.
For the remaining methods for predicting A , based on an annoyance response,
the values of AS ranged from 0 to 13 dB. The average over all the four sources and the
three remaining prediction methods (i.e., Crest Level method with or without repetition
rate adjustments, the Standard Deviation Method, and the ISO R 1996 method) is 7.2 dB.
4-6
-------
Table 6
Comparison of Several Predicted Subjective Correction Factors
for Annoyance Applied to Hie Four Impulsive Noise Sources
Correction Method
Crest Level - Value, dB
Repetition Rate -Value, Hz^ '
Munch & King (Eq. 6)
AS, dB Galloway23(Eq. 7)
Galloway (Eq. 8)
g
Peak Index
Value
V dB
Standard Deviation
Value, dB
A , dB
ISO R1996lla
A , dB
Impulsive Noise Source
Motorcycle
13
10-40(a)
0
3
2-8
-(c)
-
5
Drop
Hammer
30
0.7
13
12(b)
8
63
28
2.97
12
5
Truck-Mounted
Garbage
Compactor
19
0.2
11
6
3
21
16
1.26
5
5
Rock
Drills
19
1
11
6
3
-
1.09
4
5
(a) Assumed maximum repetition rate of 10-40 Hz for purposes of estimating
maximum A according to Eq. 8.
(b) Computed value beyond range of Crest Level for Galloway's data.
(c) Data not available for computing A .
(d) Estimated from data in Table 1, page 2-9.
4-7
-------
4. 1.4 Summary of Methods for Computing the Subjective Correction Factor A
The values for A derived from most of the preceding methods are summarized
in the following table. It was feasible to break down the comparison by the four noise
sources only for methods based on annoyance.
Table 7
Summary of Subjective Correction Factor (A )
Estimated from Existing Methods or Data, dB
Impulsive
Noise
Source
Motorcycle
Drop Hammer
Truck-Mounted
Garbage Compactor
Rock Drill
Average
Annoyance^0/
r >(b)
Crest
Level
0-3
12-13
6-11
6-11
Crest Levelfc)
Plus
Repetition Rate
2-8
8
3 '
3
Standard
Deviation
-
12
5
4
6.7±4.1
Noisiness
Fidell-
Pearsons
i:
±:
2.5
3.5
Izumi
K
±.
).5
5.5
13.0
Loudness
Tone/
Noise
±2
i
>..5
7
ISO
Round
Robin
1
±
2.5
0.9
12.5
(a) See Table 6.
(b) Based on Munch and King (Eq 6) and Galloway (Eq 7).
(c) Based on Galloway, Eq 8).
The average values for A summarized in Table 7 seem to fall into two groups.
The average values of A , based on the methods which involve direct measurement of
noisiness or loudness response with real impulsive noise sources in a laboratory setting,
are essentially identical (i.e., A$ = 12.5 to 13.5 dB). In contrast, the predicted
values of A for real impulsive sound based on annoyance criteria or measured values
of tone or noise bursts are lower (about 7 dB).
The data in Table 7 do not provide the basis for an unequivocal choice for a
means of predicting a subjective correction factor. However, the lower group of
values observed for the annoyance response has one basic point in their favor -
4-8
-------
the values are generally based on more realistic test data in terms of the test signals.
For example, the results from the helicopter tests in many cases stem from real fly-
overs for which the observed value of A was generally low or, in at least one case
24c s
(Powell ), actually negative. Thus, the actual temporal setting of the impulsive
noise signal may tend to decrease the observed value of A below that observed for
quasi-steady state sounds evaluated in a laboratory setting.
In an attempt to resolve the differences in these average results, an effort was
made to reexamine the use of Crest Level (Crest Factor in dB) to discriminate between
various degrees of impulsiveness and hence, presumably, annoyance. Thus, dis-
counting, for the moment, the negative result by several investigators regarding a
relationship between judged impulsiveness and judged annoyance, the ISO Round Robin
data were reviewed to see if such a relationship might be evident. By pooling the
information on the peak and rms values of the ISO impulsive noise samples numbers
1-9 from two specific ISO Round Robin Tests (Shipton, etal, and Thompson, etal),
it was possible to estimate the Crest Level for these sources. The subjective annoyance
correction factor A was then plotted as a function of this Crest Level. The results are
s
shown in Figure 22 along with the estimated value of A based on the methods proposed
17 23 S
for helicopters by Munch and King and Galloway and Kryter's method. For the
latter.- it was assumed that his level, Lp (i) - LpN(b), defined as the perceived
noise level of the impulse minus the perceived noise level of the background noise,
would be roughly comparable, to a first approximation, to the Crest Level as defined
in Figure 22. This comparison seems to again indicate that Crest Level alone is not
a valid basis for predicting A . The potential improvement in a prediction model for
A by including repetition rate is certainly an avenue to pursue. In any event, in the
absence of more definitive data, the following conclusions are drawn concerning an
interim method to estimate a subjective correction factor for impulsive noise sources.
1. For the type of impulsive sources of concern for this report (this excludes
helicopters), a constant subjective correction (A ) of +7 dB added to
the true A-weighted equivalent sound level for an impulsive
4-9
-------
16
14
S 12
o
O
c
O
-fc-
o
0)
1_
O
u
10
ti 6
to
O
O
O
00
Munch and King
Estimated from Kryter
7
O
O
-Galloway
O ISO Round Robin Data
77
10
15
20
Crest Level, L . (C) - L (A), dB
pk eq
Figure 22. Comparison of Measured and Estimated Values of the Subjective Correction
Factor As As a Function of Crest Level (based on C-weighted peak level
minus A-weighted rms level) (Note that Galloway defines Crest Level in
terms of the difference in Peak and rms A-weighted levels)
4-10
-------
noise source would better define its effective L , that is the L
eq eq
of an equally annoying nonimpulsive reference sound. No additional
correction is identified at this time for the possible change in A as
a function of Crest Level or repetition rate.
This first approximation leaves much to be desired in developing a more discrim-
inating correction factor. Indeed, the strong evidence of the potential validity of
improved methods for calculating loudness of impulsive sounds suggests just such an
zo
approach. According to Reichardt, improved accuracy in predicting loudness of
impulsive sounds would be provided by adding a secondary correction to the Stevens or
Zwicker loudness level equal to Aj = L. . - L. ,. where L. , and L. - are the A-weighted
"impulse" and "slow" readings taken on an impulse precision sound level meter.
Alternatively, Johnson and Robinson have computed loudness directly, using Stevens
Mark VI and a 70 ms integrating time for acquisition of spectral content data. Neither
of these approaches were able to take advantage of the latest model (Stevens Mark VII)
for loudness calculation. The need to select an optimum loudness calculation method
applicable to the type of impulsive sources considered here leads to the following
recommendation.
2. A comparative evaluation should be made of alternate forms of existing
loudness calculation methods based on either the Stevens Mark VI 1^3
87 89
or Zwicker ' models when applied to existing or new data on
subjective response to impulsive noises. Particular attention should be
given to the selection of an optimum time-constant or time-averaged
measures of level for the spectral analysis data required. Alternatively,
the methods proposed by Izumi may offer an improved procedure for
predicting A and should be explored further.
It is anticipated that values of A computed with such improved models will
show more discrimination as to the magnitude of A versus one or more signature
characteristics of the impulsive source such as Crest Level or peak to background
noise levels. A valid data base for computing the subjective correction factor for
any one category of impulsive noise sources is required. This data base of one-third
octave spectra must be acquired for a sufficient number of units to ensure a valid
sample of the total population.
4-11
-------
Finally, the evidence that repetition rate is potentially significant in the
development of valid subjective correction factors leads to the final recommendation:
3. Further research is needed to explore, in more detail, the significance
of repetition rate on the subjective response to impulsive sounds. This
research should also consider the potential need to extend or refine
estimates of loudness or noisiness contours to lower frequencies where
spectral peaks due to repetition rate may be significant.
Other areas for improvement in understanding subjective response to impulsive
noise also exist. These include such areas as developing a better understanding of
hearing damage risk to impulsive sound, correlating annoyance versus loudness or
noisiness responses and evaluating sleep disturbance due to impulsive noise. This
report has attempted to provide an overview of most of these problems and, hopefully,
provide a basis for practical steps to be taken now for evaluating the environmental*
impact of impulsive noise sources. Problems related to the objective measurement of
such sounds are addressed in Appendix A to this report.
4-12
-------
APPENDIX A
OBJECTIVE MEASUREMENT OF IMPULSIVE NOISE
A. 1 Introduction
A measurement of impulsive sounds which accurately represents their annoying
quality has clearly presented a major challenge to acousticians. This difficulty is
mainly derived from the inability of a piece of electronic hardware to faithfully reflect
the way the human ear detects, processes, averages, interprets, stores and finally
discards complex incoming acoustical signals of widely varying physical parameters.
An impulse precision sound level meter (ISLM) which attempts to approach this ideal
in one instrument has several meter settings: PEAK HOLD, IMPULSE, FAST, and SLOW
which can be combined with the various weighting networks: A, B, C, D, and LINEAR.
Any one of these settings can be applied to only a limited range of physical parameters.
For impulsive sounds, for instance, the reading cannot be expected to be within the
accuracy limit of the instrument unless the characteristic period of the impulsive sound
is substantially greater than the overall response time-constant of the electronics for
that particular setting. Thus, the task of monitoring impulsive sounds involves both the
problem of finding a procedure which will accurately reflect the physical phenomena,
as well as the even more difficult problem discussed in Section 3 of predicting the
subjective response to impulsive sounds. We shall be concerned in this Appendix with
me first task - the measurement problem for which References 131-148 are pertinent.
The goal for evaluation of the objective correction factor was to define the
difference between the true and measured L * for a variety of impulsive sounds. Based
eq
on laboratory experimentation, the A-weighted SLOW meter setting was selected to most
closely approximate the L of impulsive sounds. An objective correction factor is men
defined to add to readings from this meter setting to give the corrected baseline metric,
'See paqe 2- 10 in Section 2 for definition of L
r eq
A-l
-------
viz. A-weighted L . Various physical parameters of the impulsive sounds, such as
Hie crest level, pulse duration, period, spectrum content, and rise time of impulsive
sounds are presumably parametric to the objective correction factor. The difference
between the calculated L and the A-SLOW meter reading will be plotted against
eq
various important physical parameters of the signals.
A.2 Current State-of-the-Art of Impulsive Noise Measurement
With laboratory measurement of impulsive noise, using sophisticated electronic
equipment, a majority of the important physical parameters of an impulsive signal can
be studied in detail. This provides a more accurate evaluation of impulsive noises
than analyses made in the field with simple equipment. However, if on-site evaluation
is required, the measuring instrument must be portable, compact and easy-to-operate.
The Impulse Sound Level Meter is such an instrument. Although it is constructed in
conformance with established standards, it may give only a crude assessment of the
annoying quality of the noise.
A.2.1 Laboratory Methods
• Time History
One of the most powerful tools in a laboratory for investigation of transient
signals is the Cathode Ray Tube (CRT) Oscilloscope. With it the time history of the
instantaneous sound pressure can be displayed visually and a photograph taken for a
permanent record. Such photographs were shown in the main body of this report (see
Figure 2) for noise from a two-stroke motorcycle, a pavement breaker, a rock drill,
and a commercial truck-mounted garbage compactor. The rise time, amplitude, duration,
and period of impulsive noise are easily read using this method. If a significant pure
tone is present in the noise, its frequency can also be estimated.
If the time history of the detected level is of interest, a high speed graphic
level recorder can be employed with suitable writing and paper speed settings to
measure the envelope of the rms or peak value of the instantaneous level. Figure A-l
A-2
-------
Figure A-l. (a) Time History of a Two-Sh-oke Motorcycle
Figure A-l. (b) Time History of a Drop Hammer
A-3
-------
HI _ _i .—rrr~
I - - - - I
'.-:-- 0-:-" 1 sec 2 sec
i"l
Figure A-l. (c) Time History of a Rock Drill
" t '~~T
Figure A-l. (d) Time History of a Commercial Trash Truck with Compactor
A-4
-------
shows the time histories of the rms magnitude of the four impulsive sound sources men-
tioned above. For very short duration impulsive sounds, an oscilloscope driven by a
log amplifier can also be used to portray the time history of the signal envelope.
• Spectrum Analysis
Another powerful tool in the laboratory is the Real Time Analyzer (RTA) which
can be used to determine the detailed spectral content of impulsive sounds over the
audible frequency range. Figure A-2 shows the frequency spectrum from the four
impulsive sound sources mentioned above. However, the spectral analysis measurements
of short transient sounds is subject to appreciable error unless due consideration is given
to the transient response of the filters and to the use of an adequate integration time-
142
constant.
• Digital Analysis and Computation
A Fourier analyzer coupled to a high-speed digitizer and an electronic computer
provides the most powerful, state-of-the-art approach for analysis of impulsive sounds.
A.2.2 Field Method
• Sound Level Meter (SLM)
Generally, the SLM is designed to conform with one or more internationally-
recognized standards. Therefore, the built-in specifications for any given SLM will
nor vary significantly from manufacturer to manufacturer. Thus, an important general
observation may be made regarding the four RC-inregrating and averaging time-constants
137
in the so-called "Impulse Sound Level Meter (ISLM). "
At the "PEAK HOLD" position, the RC-network has a time-constant of 50 MS.
At settings of IMPULSE, FAST, and SLOW, the nominal effective time-constants are
136 137
35 ms, 125 ms, and 1 sec respectively. ' These time-constants, in general, do not
include the magneto-mechanical inertia effect of the analog indicating device which tends to
A-5
-------
CO
"a
Third Octave Band 20 40 60 160 3I5 630 1.25 2.5 5 10 20 40 Lia
Center Frequency 16 _315 63 ...125 __250 ...50O ...1kHz .2 .4 .8 ...16 __315 Weight
125 25
50
100
200 400
800
1.6
3.15 63 125 25(i«iMiu D A B C Un)
Figure A-2. (a) One-Third Octave Spectrum of a Two-Stroke Moforcycle
CO
"a
Third Octave Band
Center Frequency"
20| 40
315 63
25 50 100
160 315 630 1.25 Z5
126 250 500 1kHz 2 4
200 400 800 14 3.15
8 16" 315 -._._-._
&3 126 25[Miwm (5 A 8 CUU
Figure A-2. (b) One-Third Octave Spectrum of a Drop Hammer
A-6
-------
oo
K
J
5
_o
Third Octave Band
Contof Frequency
-
I 777
:-E
-77
:~
—
—
2.5 1
—
777
^
77
J5
—
77
—
2
6
~_
-
--
77
•.77
"r
0
2
77]
.777
£
-
:
_
...
3
5
i
.5
*
t
:
.
4
—
df
77
7
-1
1
'-
-•
0
5
-
-
—
—
—
-_-_
-'•
6
0
^
-
77
7
--
8
3
—
—
777
-r
-I
77.
--
~
—
0
1(
T:
—
777
J
777
-_
77
777
-
i:
)0
—
~
__.
77
__
~
—
—
If
>5
._
7777
77
.!»
—
—
77
77
—
>0
«
_..
77
—
77
77
Z
»
—
3-
=
—
—
"77
-
.17
~
31
«
77
_7
"1
1
~
-
-;
-
-
5
4,"
'—
--
__.
|
-
._ .
-
--
5C
O
-
:
K>
— i
:
-
Gl
777
77
—
r--
£-
-
--"
—
u
8C
—
-
—
—
--
-
••* -
.7.
77
Ik
K)
--
:-
--
—
•"-
"~
1.
Hz
-
--
—
77
i
—
-
777
Ib
1
—
—
—
—
—
-r
—
— "
77
:
6
—
—
: 1
-r-
=f
~
—
~~
•L
—
-.-
••v-
-
~-
*J
3.
___
~:
77
4
5
_...
-
I
"--
:-.
_:
._
—
i—
=
—
77_
r-
—
6
—
-
z:
-
•"
e
3
—
—
Tl
id-
_
"--
.7;
1
—
77-
T
:r-
—
~_
u
12
-
.-r
777
77
—
t
5
—
—
_ _
—
[
L
-
—
'2
6
__
U
_
••
y
--
—
—
^
777
-~
r.-
_-
—
31
—
—
—
77:
—
-
—
4(
5
-
-
_
J
N
-M-
".
J--
77 '.
L
. J.
_
77.
Lin
-•nhL
D A
-f-H-
i i i
i i i
i
i
i i
1 {__[_
-1 h
T"
-j^$
71777 777T-
"i — — r
—
B CUnT)
Figure A-2. (c) One-Third Ocfave Spectrun of a Rock Drill
CO
—
>
J3 f
t>
*
Trwd Octam Band
a
3
--
-0
1
4
1 —
-4
--
3
i
— —
T :
—
~|.4 — I
1 ..
80
1 [
\ 1
T-
• V.
160
— —
-1 1
•Jb-77
31
—
—
-
77
5
77
.7
—
63
.IT .
to
—
Jj
4 -I
—
—
i
1 — I
i.
—
—
25
1
a
N
5
-t
^~
i
-H
i
0
r
•j
u
—
—
4
u
-
0
1-
J
t
7
i
_
j_
_
+-
1
1
1
1
j
1
r
i
i
^
_i
— }-
~ r
i
L
1
__*...
1
(
-t-
i
i
i
i
i
_L
1
1
Ua
J-f-+
i i
i i
i I
i 1
i i
i i
J 1 1~
I
i
1
i
i
i
i i
TJTTTl-T
CentarFnequenqr^TS—^U ^ 63 ,^125 ^250 ^500 ^IkH, ^ S ^4 ^ - ^^ -B,,^,, ,5 A B c ^
Figure A-2. (d) One-Third Ocrave Spectrum of a Commercial Truck-Mounted
Garbage Compactor
A-7
-------
increase Hie overall end-to-end time-constants of the ISLM.* It is possible to roughly
identify the characteristic period of repetitive impulsive sounds according to the degree
of fluctuation which occurs in the read-out device as a result of these different time-
constants. For instance, with a SLOW meter setting and an impulsive signal of very
short duration and period, the amount of signal charging and discharging through the
integrating capacitor on the output of the meter detector would be minimal, resulting
in a steady reading of the ISLM. For an impulsive signal of very long duration and
period, on the other hand, the capacitor is fully charged and discharged during
each cycle and a large fluctuation in the meter reading would result. Therefore,
from the degree of the fluctuation of the reading, a rough idea of the combined
duration and period of the impulse can be roughly estimated (see Figure A-3).
In addition to the effect of internal time-constants of the ISLM, another
important parameter which reduces the accuracy of the ISLM is the crest level of the
input signal. By carefully adjusting the position of the ISLM input and output attenu-
ators to avoid saturating the amplifier of the ISLM, the reading accuracy can be
improved. However, the inherent uncertainty in the meter reading for maximum crest
141
level signals that can be handled by the ISLM is approximately ± 1 dB. The orien-
tation of the ISLM with respect to the impulsive sound source, the distance from it,
and the general physical environment surrounding the sound source will also influence
the reading obtained from the ISLM.
• Spectrum Analysis
An octave band filter can be used in conjunction with the ISLM to determine
the approximate frequency distribution of an impulsive noise; however, the accuracy
is necessarily limited by the transient response characteristics of the filter.
*A decay time constant of 3 sec is provided for the ISLM to partially compensate for
this meter slugglishness.
A-8
-------
CO
0)
O
1 second
T > 1 second
^_ Large Meter
Fluctuations
Time Scale
Figure A-3. Fluctuation of the Sound Level Meter Readings Versus Pu Period
and Duration
A-9
-------
A.3 Experimental Procedure
In order to systematically investigate the output of an Impulse Sound Level
Meter due to impulsive signals, a wide range of synthesized signals was used to cover
the three regions of meter fluctuation. Readings,fluctuated the least, of course, for
the SLOW meter setting. The most steady conditions were obtained for impulse durations
less than 100 ms and periods less than 1000 ms (see Figure A-3). Thus, the SLOW
meter setting was used to define an objective correction factor, A , for impulsive
sounds as follows:
A = L - LAC (A-l)
o eq AS
where LA <• is taken to be the maximum reading of the fluctuating meter needle afid
L is the equivalent sound level based on the duty cycle and sound levels of the tone-
burst and background noise. The variation of A with respect to various physical param-
o
eters of impulsive signals was fhen examined. Since the A-weighted equivalent sound
level (L ) was selected as the baseline metric, the A-weighting network was chosen
eq
in conjunction with the SLOW meter setting to read impulsive sounds in order to mini-
mize the variation of the objective correction factor as much as possible. However,
the objective correction factor was expected to be meaningful only for impulsive
signals which produced a small meter fluctuation or produced a definitive trend of
A based on the maximum meter reading for signals with larger fluctuation (see Figure
A-3).
The reason for choosing the maximum meter reading for signals with other than
small fluctuation is based on the fact that any fluctuation of more than 10 dB will be
difficult to observe with the same attenuator settings of the ISLM. Consequently, only
either the maximum or the minimum reading can be read at any one time, and the
maximum value was considered much more informative.
The physical parameters of the impulsive signals used in this experiment are
listed in Table A-l. Three types of background noise are used: none, pink, and
USASI.149
A-10
-------
Table A-1
Range of Physical Parameters of Hie Synthesized
Impulsive Signals Used in This Study
Duration
.4-400 ms
Period
2-4000 ms
Frequency
20 Hz-10 kHz
Crest Level
15-35dB
Signal-ro-Noise Ratio, dB
10-50
A block diagram of the instrumentation is shown in Figure A-4. The measure-
ment procedure is as follows: The level of the background noise is first set to a given
SPL as read by the ISLM.* With the background noise off, a continuous sinusoidal
signal of a given frequency is similarly set to a different level to provide a given "signal-
to-noise" ratio. This continuous signal is then changed in temporal pattern only to a
tone-burst with a preset duration and period which is then superimposed on top of the
background noise. Finally, the combined signal is fed into the ISLM and readings are
taken.
A.4 Results
All the observed and computed data have been tabulated in Table A-2. The
computations for the values in columns 10 and 11 are explained in the footnotes at the
end of the table. The master index in Table A-2 is the pulse duration (PD), given in
the first column, which ranges between 0.4 ms to 400 ms. The next sorting is on the
period (T) in the second column, which varies from 2 ms to 4 seconds. The duty cycle
is not listed, but is equal to 100 (PD)/T, %. It varies from 0.1 percent to 50 percent.
The next sorting is on the center frequency of the tone burst in the third column, which
ranged from 20 to 10,000 Hz. The final sorting was usually on the signal-to-noise ratio,
defined in the table, and listed in column 10, which varied from 5 dB to 50 dB. The
measured crest level, which is defined as L - L , varies from 10 dB to 35 dB. The
pk S
range of these parameters is considered large enough to embrace most of the impulsive sounds
*The ISLM (B& K Model 2204/S) performed according to the manufacturer's
specifications on single and repeated tone bursts.
A-ll
-------
Frequency
Counter
Noise
Generator
Frequency
Generator
Tone- Burst
Generator
A A A A (
V V V '
Cathode
Ray Tube
(CRT)
Impulse
Sound Level
Meter (ISLM)
Figure A-4. Block Diagram of the Instrumentation Used in Generating and Measuring Objective
Correction Factors for Artificial Impulsive Sounds
-------
which are of particular interest. The objective correction factor, A = L - LA r, is
o eq AS'
listed in the fourth column.
In studying the table, several points are of interest. First, note that when
L^p starts to fluctuate for a particular pulse duration and period, L._ remains steady.*
The second point to be noted is that when both the duration and period become longer,
L. _ starts to fluctuate also. The greatest fluctuation in L._ occurs when the pulse
duration (PD) is on the order of 100 ms and the period exceeds 2 seconds. Impulsive
signals for which the duration is over 500 ms and the period is over several seconds
have not been included in the measurements since no real sounds which were analyzed
fall into this range.
In Figure A-5 the objective correction factor, A = L - LA c, has been
o ecj r\ o
plotted for a constant frequency and pulse duration against the measured crest factor
L , - L_ for several values of the duty cycle. The correction factor remained nearly
constant in the range of 0 to +2 dB. The average objective correction factor is 0.78
dB with a standard deviation of 0.45 dB. The correction factor is plotted against
S/N in Figure A-6. The scattering of the data points is small, but no definitive trend
with varying parameters was observed. The mean and standard deviation of A is given
on the figure. In the plot of A versus frequency (Figure A-7), the data scatter has
increased but still no definitive trend resulted. From Figures A-8 to A-13, the objec-
tive correction factor has been plotted versus period (T) (for constant pulse duration),
pulse duration (PD) (for constant period), and duty cycle, for impulsive signals with
Uttle or no fluctuation in the SLM (Figures A-8 to A-10) and for signals which cause
substantial fluctuation in the SLM reading (Figures A-ll to A-13). For the latter, the
data are based on the maximum meter reading. The scatter of the data ranges between
+2.0 dB to -3.0 dB. A gross downward trend is evident in Figures A-8, A-9 and A-12.
Although this trend is not clearly defined by the data, it would seem to suggest a signifi-
cant decrease in the average value of A (accompanied by an increase in data scatter)
when the pulse duration substantially exceeds 100 ms or the period exceeds 1 second.
*Significant meter fluctuation for any condition is signified in Table A-2 by two values
for the SLM reading (i.e., the maximum/minimum reading).
A-13
-------
Table A-2
Summary of Experimental Data Obtained from Laboratory Synthesized Impulsive Sounds
PD(ms)(1)
.4
1
T
5000
1000
1
Leq - LAS
+1.2
-0.4
0.7
0.8
0.8
-0.5
0.5
0.6
-0.2
1.7
1.0
1.0
1.1
1.0
0.4
0.8
0.4
0.6
0.6
0.5
0.5
0.8
0
-0.1
L (4)
eq
115.5
89.1
92.8
97.3
102.1
86.2
88.0
91.2
95.4
82.9
86.8
95.2
105.0
115.0
82.5
85.1
92.4
102.0
112.0
82.3
83.5
90.0
98.1
108.0
L (5)
LAS
114.3
89.5
92.1
96.5
101.3
86.7
87.5
90.6
95.6
81.2
85.8
94.2
103.9
114.0
82.1
84.3
92.0
101.4
111.4
81.8
83.0
89.2
98.1
108.1
4<"
113.5
83.5
86.5
95.2
104.8
114.8
84.5
86.2
93.0
102.4
112.3
84.3
85.2
90.5
99.2
109.1
L (/)
AF
114.3
88.8/89.4**
92.2/92.0
96.6/96.4
101.4/101.2
86.8/86.4
87.7/87.3
90.8/90.5
95.6/95.4
L (8)
LAI
117.3
89.9/89.5
92.4/92.1
96.7/96.6
101.6/101.3
37.3/86.9
88.3/87.7
91.4/91.3
96.2/95.9
^
pk
137.2
102.0
109.2
117.8
128.6
137.6
102.0
109.4
117.8
128.8
137.5
101.4
109.2
117.8
128.7
137.5
S/N<10>
50
1.5
6.5
11.5
16.5
1.5
6.5
11.5
16.5
10
20
30
40
50
10
20
30
40
50
10
20
30
40
50
C.L.(11)
-23.7
18.5
22.7
22.6
23.8
22.8
17.5
23.2
24.8
26.4
25.2
17.1
24.0
27.3
29.5
28.4
Remark
U.A. 82.1 dB*
P. A. 85 dB***
1
U.A. 82.1 dB
•
>
I
-------
Table A-2 (Continued)
PD ms/1^
1
2
r
T,™,<2>
1000
4
20
'
200
2000
1
F (Hz/3'
1000
500
i
Leq - LAS
0.7
0.9
0.8
1.2
1.6
0.5
0.7
0.8
0.9
0.5
0.6
0.9
0.3
-2.0
+0.6
+0.4
+0.5
+.5/-1.4
+.2/-S.7
-.3/+»t
-.7/+»
+0/+.5
+.5/+3.S
-.3A-
eq
82.2
82.9
86.8
95.2
105.0
89.1
92.8
97.3
102.1
86.2
88.0
91.2
95.4
81.8
92.2
101.8
111.8
81.5
84.7
92.2
101.8
82.5
85.0
92.2
LAS(5)
81.5
82.0
86.0
94.0
103.4
88.6
92.1
96.5
101.2
85.7
87.4
90.3
95.1
83.8
91.6
101.4
111.3
81.0/80.1
84.5/79.0
92.5/<90
102.5/<100
82.5/82.0
85.5/81.2
92.5/<-»tt
4(6>
84.3
84.5
87.8
95.0
104.2
87.6
95.4
105.2
115.2
84.4/83.5
88.5/82.5
96.8/<90
106.5
85.5/84.8
89.0/84.0
96.5/<90
LAF(7)
88.8/88.4
92.2/91.9
96.6/96.4
102.2/101.1
85.6/85.9
87.5/87.2
90.4/90.1
95.0/95.2
LAI(8)
89.0/88.7
92.4/92.1
96.8/96.6
104.4/101.3
86.4/85.9
88.4/88.0
91.5/91.0
95.9/96.1
v(9)
101.3
109.2
117.6
128.6
137.5
109.4
118.6
128.5
137.6
109.4
118.6
128.6
137.4
108.5
117.3
127.4
S/N<'°>
10
20
30
40
50
4.7
9.7
14.7
19.7
4.7
9.7
14.7
19.7
8.2
30
40
50
0
30
40
50
20
30
40
c.J»)
17.0
24.7
29.8
33.6
33.3
21.8
23.2
23.3
22.4
25.0
30.1
31.8
30.9
23.0
28.3
30.9
Remark
U.A. 82.1 dB
P. A. 85 dB
P. A. 81.5dB
U.A. 82.1 dB
Ui
-------
Table A-2 (ConMnued)
PD(ms)(1)
2
4
5
T, .(2)
T (ms)
2000
400
!
10
r
25
50
100
F(Hz)(3)
500
500
1000
5000
200
1
j
T
400
1000
L - L
-.2/+-
-.S/-.4
-.S/-.7
-.9/-.S
0.4
0.5
0.5
1.0
1.2
1.2
0.9
1.0
1.1
0.7
0.9
1.0
1.1
0.6
0.7
0.7
-1.3
-.2/.1
0.6
eq
101.8
111.8
115.0
115.5
91.6
101.2
111.1
88.3
97.2
107.0
86.2
94.1
104.0
86.2
88.0
91.2
95.4
84.6
91.5
101.0
71.1
77.2
82.0
L (5)
AS
102.0/<-»
112.3/112.2
115.8/115.8
116.4/116.3
91.2
100.7
110.6
87.3
96.0
105.8
85.3
93.1
102.9
85.5
87.1
90.2
93.3
84.0
90.8
100.3
72.4
77.4/77.3
81.4
, (6)
S
1 06.3/0"
115.9/0.8
116.0/115.9
115.6/115.5
102.0
111.8
121.8
97,4
107.2
117.1
94.7
104.2
114.2
92.3
101.7
111.6
82.1
82.5/82.3
81.7
LAF(7)
112.2/109.5
115.6/112.9
116.3/113.0
85.7/85.3
87.2/86.9
90.3/90.1
93.2/93.5
72.3/72.1
77.3/77.0
81.3/81.1
LAI(8)
121.4/121.3
125.2/125.0
125.7/125.5
86.7/86.4
89.1/88.8
92.6/92.4
96.6/96.3
77.1
82.9/82.8
87.2/86.6
LPk(9)
137,4
137.7
137.5
137.2
110.4
119.2
128.2
110.2
119.3
128.1
110.3
119.3
128.1
110.2
119.2
127.4
98.5
98.3
98.7
S/N(10)
50
50
50
50
20
30
40
20
30
40
20
30
40
20
22.6
26.5
31.1
20
30
40
c
o
C.L.
31.1
21.8
21.5
21.6
8.4
7.4
6.4
12.8
12.1
11.0
15.6
15.1
13.9
17.9
17.5
15.8
16.4
15.8
17.0
Remark
U.A. 82.1 dB
No BNttt
-------
Table A-2 (Continued)
PD(ms)(1)
5
8
10
20
i
T(»,P>
100
'
250
800
20
100
IOC
,
0
200
I
1
F (HZ)(3>
2000
4000
10,000
200
500
1000
5000
100
5C
1
)0
1000
5000
50
•
Leq - LAS
0.6
0.6
0.8
0.2
-0.4
-0.6
1.1/1.6
0.1/0.5
-.2/.4
0.9
0.1
-0.3
-0.3
0.4
0.3
0.1
1.0
-.7/.1
-.S/.4
-.4/.6
-.08
-1.7
-2.4
L
eq
83.2
83.0
79.5
83.3
88.2
97.1
118.8
115.0
115.5
89.1
92.8
97.3
102.1
86.2
88.0
91.2
95.4
96.1
111.8
115.0
115.5
86.2
88.0
91.2
LAS(5)
82.6
82.4
78.7
83.1
88.6
97.7
111.7/111.2
115.1/114.5
115.7/115.1
88.2
92.9
97.6
102.4
85.8
87.7
91.1
94.4
98.0
112.5/111.7
115.5/114.6
115.4/114.9
87.0
89.7
93.6
4'"
81.6
81.6
81.4
90.3
99.3
108.0
115.1/114.5
115.1/114.6
115.0/114.5
118.2/117.2
115.8/115.0
115.0/114.6
115.2/114.2
I (7)
LAF
82.4/82.3
82.2/82.1
78.5/78.3
114.8/-"
118.3/-=o
119.4/—
88.3/88.0
93.0/92.8
97.7/97.5
102.4/102.3
85.9/85.6
87.8/87.5
91.1/90.8
94.5/94/2
103.5/-»
120.2/-«
120.4/-»
87.1/86.8
89.8/89.4
93.5/93.1
LA/8)
87.6
87.4
83.6
124.3/123.0
128.0/126.5
128.8/127.3
89.0/88.7
93.5/93.3
98.2/98.1
102.9/102.8
87.7/87,4
90.6/90.2
95.0/94.7
99.3/99.1
111.9/110.0
126.1/124.0
128.7/126.7
129.2/127.0
90.0/89.7
94.2/94.0
98.3/98.0
S*1"
98.5
98.3
98.1
110.0
119.4
128.0
137.6
137.5
137.3
137.9
137.9
137.4
137.3
s/N'10'
CO
20
30
40
50
50
50
20.6
25.6
30.6
35.6
20
25
30
35
50
50
50
50
20
25
30
.6
.6
.6
.6
.6
.6
.6
C.L<">
16.9
16.7
16.7
19.7
20.1
20.0
22.5
22.4
22.3
19.7
22.1
22.4
22.1
Remark
NoBN
U.A.
82.1 dB
P. A.
85 dB
U.A.
P. A.
8
2.1 dB
85 dB
-------
Table A-2 (Continued)
PD(ms)(1)
20
50
T (™P
2000
500
1000
F (HZ)(3)
50
100
500
1000
5000
50
1000
<
L - LAC
eq AS
1.2/1.3
1.6/2.6
-4.8
-3.9
-2.7
-2.8
-3.2
-fO.8
+1.8
-2.4
-1.1
-1.2
-0.8
-0.8
-1.3
-.3/+.4
-.4/+.S
-.S/+.6
-.3A.5
L «>
eq
81.5
81.6
82.3
85.9
86.7
96.1
111.8
115.0
115.5
82.1
82.2
82.8
86.2
86.2
88.0
95.2
105.0
115.0
125.0
92.4
102.0
112.0
122.0
L (5)
AS
80.3/80.2
80.0/79.0
****
****
91.5
100.0/-»
114.5A100
117.8/-100
118.7/~100
81.3
81.0
****
****
88.6
89.1
96.4
105.8
115.8
126.3
92.7/92.0
102.4/101.5
112.3/111.4
122.3/121.5
L (6)
89.0/83.0
97.1/84.0
106.8/91.0
116.8/100.5
120.3/100.0
119.2/-100
117.8/-100
117.8/-100
117.9/-100
89.0/83.0
97.2/84.0
106.8/90.0
116.7/110.0
96.6
105.9
115.8
126.3
93.1/92.3
102.4/101.5
112.3/111.4
122.4/121.5
L <7>
AF
****
105.5/-"
120.3/-«
123.4/-»
124.0/-"
89.6/86.2
90.0/87.0
I (8)
AI
103.0/98.0
113.2/106.2
128.5/122.5
131.6/126.0
132.2/126.6
94.8/94.3
95.1/94.6
L <9>
V
109.4
117.8
127.7
137.8
137.9
137.8
137.4
137.5
137.2
109.3
118.0
127.8
137.8
110.5
119.3
128.9
137.9
110.2
119.2
128.9
137.8
S/N<10>
40
30
40
50
50
50
50
50
50
20
30
40
50
15
20
20
30
40
50
20
20
40
50
C.L.*1"
20.4
20.7
20.9
21.0
17.6
18.6
19.6
19.7
19.3
20.3
20.8
21.0
21.1
13.9
13.4
13.1
11.6
17.1
16.8
16.6
15.4
Remark
P. A. 81.5dB
U.A. 82.1 dB
P. A. 85 dB
1
U.A. 82.1 dB
•
I
00
-------
Table A-2 (Continued)
PD (ms)(1)
50
f
100
i
200
!
400
I
T W2>
2500
i
1000
T
2000
1000
2000
T
800
1
F (Hz/3)
1000
T
20
'
1000
1
200
400
1000
2000
4000
10,000
20
Leq - LAS
-1.7/+"
-3.7/+»
-3.8/+»
-4.0/+ <=
-.7/-1.0
-5.2
-2.7/+12.2
-2.8/+»
-2.7/+ <»
-2.8A-
-2.5/+»
-2.3/+°
-2.5A"
-2.6/+co
-2.5/+°
-2.5/+CO
-1.2/.-6
-1.7A1.7
-0.9
-0.2/0.1
L (4)
eq
89.6
98.1
108.0
118.0
86.2
88.0
92.4
102.0
112.0
122.0
71.1
77.2
82.0
83.2
83.0
79.5
87.1
86.2
88.0
89.1
92.8
L (5)
LAS
92.3A80
101.8A80
1 1 1 . 8/< 90
122.0/2 100
87.5/87.2
93.2
95.1/80.2
104.8A90
114.7A 100
L,">
92.6/82.7
102.1 /< 80
1 1 1 . 8/< 90
122. OA 100
95.2/82.8
104.8A90
114.7A 100
124. 8/< 100 124. 8/< 100
73.67—
79.57—
84.57—
85.8A"
85.57—
82.0/—
88.3/87.8
87.9/84.5
89.4/85.0
90.0
93.0/92.7
84.7/-»
84.57—
84.5/-<=
84.5/-»
84.5/-»
84.3/-»
L (7)
LAF
90.6/88.2
93.4/92.7
79.4/-»
85.3/-<=
90.3/-»
91.5/-»
9 1 . 3/-»
87.8/-»
91.4/83.0
91.9/83.0
94.0/84.0
91.4/86.8
95.0/87.5
L (8)
LAI
94.9/92.9
97.0/96.8
83.5/77.5
89.5/83.0
94.5/88.5
95.8/90.0
95.6/89.8
92.1/86.2
94.8/93.0
95.1/90.3
96.8/92.0
94.6/92.3
96.8/95.4
L (9)
Lpk
110.1
119.2
128.7
137.7
109.7
119.2
128.8
137.8
99.1
99.2
99.2
99.0
99.3
98.8
S/N<10>
20
30
40
50
52
57
20
30
40
50
c
D
52
52
57
52
57
C.L
17.5
17.1
16.9
15.7
14.5
14.4
14.1
13.0
14.4
14.7
14.7
14.5
14.8
14.5
Remark
U.A. 82.1 dB
P. A
U.A
. 85 dB
T
. 82.1 dB
I
No BN
P. A
85 dB
-------
Table A-2 (Concluded)
PD (ms)(1)
400
T (msP
2000
I
F (Hz/3)
20
i
I
1
L - L.c
eq AS
-1.8/+1.8
-2.27+2. 5
-3.0/+3.2
-3.6/+4.0
L <4>
eq
87.1
88.8
86.2
88.0
L (5)
AS
88.9/85.3
91.2/86.3
89.2/83.0
91.6/84.0
s(i)
92.0/83.6
94.8/84.0
92.0/83.0
94.8/83.5
L (7)
AF
94.7/90.9
96.8/93.0
94.8/83.0
97.1/85.0
L (8)
AI
L,(9)
pk
S/N<10>
52
55.4
52
57
C.L.
Remark
P. A. 85 dB
>
(2).
'PD:
V:
F:
L
eq
(5),
(\r
(8), .
LAP
(9)L .
V
<10>S/N:
T
tt
TTT
Pulse duration, ms
Period of the impulse train, ms
Center frequency of the synthesized pulse, Hz
Computed A-weighted equivalent continuous sound pressure level, dB
Impulse Sound Level Meter reading at "SLOW" meter setting with A-weighting network, dB
1SLM reading at "SLOW" with no weighting, dB
ISLM reading at "FAST" with A-weighting network, dB
ISLM reading at "IMPULSE" with A-weighting network, dB
ISLM reading at "PEAK HOLD" with flat weighting, dB
Difference between the unweighted rms levels of the background noise and the continuous sinusoidal tone (prior to tone bursting), dB
Crest Level. Defined as L . - L,, dB
pk S
Background noise, USASI, A-weighted (U.A.), 82.1 dB (see Reference 148 for description of spectrum)
A virgule separates upper and lower readings from the same meter setting
Background noise, pink, A-weighted (P.A.), 85 dB
+ =• means that the difference is more than 10 dB
-™ means that the SLM reading is too small to be registered for a particular setting of attenuators
No background noise (BN)
-------
6.0
4.0
2.0
CO
,< o
cr
-2.0
-4.0
-6.0
Frequency = 1000 Hz
Pulse DuraHon = 1 ms
Duty Cycle
O 0. 1 Percent
O 0.2 Percent
D 0.5 Percent
A 1 Percent
Average = .78 dB
S.D.= .45 dB
10
AAA
20
L k - L (Crest Level ), dB
30
o
40
Figure A-5. Correction Factor fora Constant Frequency and Pulse Duration Versus the Measured Crest Level
-------
NJ
S3
6
4
2
n
TD
CO
_,< o
1
cr
o -2
<3
-4
-6
Average = .45 dB
S.D. = .68 dB
-
a
1
— Frequency Duration
Hz ms
a 1000 1
H
A 500 2
o 200 5
G
A
0
o 50 20
20 50
i i
i " — i ' f '- 1"11 - i i i
-
-
A
63 JdA _{|
Period A • — .
ms * 0* ? *"
X X
100 x -
200
500
1000
2000
10
25
50
100
250
2000
500
1000
i 1.1 1 i i i
10
20
30
40
S/N, dB
Figure A-6. Correction Factor Versus Signal-to-Noise Ratio for Steady Readings
50
-------
03
-o
I/O
*
Duration (ms) Period (ms)
A
D
O
O
•
5
8
10
20
100
50
50
20
100
800
1000
2000
2000
500
1000
200
Average = -1.29 dB
o
-2
-3
-4
10
100
1000
10,000
Frequency, Hz
Figure A-7. Correction Factor Versus Pulse Frequency for Steady Readings
A-23
-------
Average = .05 dB
S.D. = 1.10 dB
<
-1
-2
-3
-4
- o
D
D
A
A
O
A
Duration (ms)
0.4
.0
.0
0
.0
1
2.
4.
5.
8.0
10.0
20.0
50.0
i
a
A
AA
I!
A
I
O
O
D
O
i
A
A
A
O
O
10
TOO
T, Period, ms
1000
Figure A-8. Correction Factor Versus Period for Steady Readings
-------
3 ]
< o
-2 -
-3 -
-4 -
Average = . 03 dB
S.D. = 1.25 dB
o
o
•
Period (ms)
40
100
200
400
500
800
1000
2000
J
0.1
1
10
100
Figure A-9.
PD, Duration, ms
Correction Factor Versus Impulse Duration for Steady Readings
-------
3
2
1 j-
0.1
Average =0.29 dB
S.D. =0.86 dB
co
t)
*,
<
»J
u-
_J
II
o
<
0
-1
-2
-3
-4
»i
** I * r
•* • • «
•
• •
• •
•
•
• •
•
»
•
•
N *
•
•
-
—
1 1 1
1 10
Duty Cycle, Percent
Figure A-10. Correction Factor Versus Duty Cycle for Steady Readings
100
-------
Average =
S.D. =
-1.42 dB
1.21 dB
o
o
co
<
cr
0)
o
-1
-2
-3
-4
A
O
O
A
a
a
Duration
2
10
20
50
100
200
400
A
a
8
8
i
100
1000 10,000
T, Period, ms
Figure A-l 1. Correction Factor Versus Period for Fluctuating Readings
-------
Average = -1.31 dB
S.D. = 1.42 dB
CO
-o
-1
-2
-3
-4 -
A
A
A
Period (ms)
a 1000
A 2000
o 2500
I
a
D
A
A
10
o
100
A
PD, Duration, ms
Figure A-12. Correction Factor Versus Duration for Fluctuating Readings
A
A
1000
-------
0
-2 -
-3
Average = -1 .45 dB
S.D. = 1.41 dB
-4
0.1
.0
10
100
Duty Cycle, Percent
Figure A-13. Correction Factor Versus Duty Cycle for Fluctuating Readings
-------
A. 5 Conclusions - Objective Measurement of Impulsive Sounds
The ISLM readings have been divided into two categories: those obtained from
the signals with a repetition rate greater than 1 pulse per second (pps) small fluc-
tuation region and those with a repetition rate from 0.4 to 1 pps. The average value
and standard deviation of the objective correction factor for these two regions are:
Average A Standard Deviation
o
When repetition rate > 1 pps +0. 1 dB 1.3 dB
When repetition rate = 0.4 to 1 pps - 1.4 dB 1.4 dB
The objective correction factor has also been studied for various temporal
parameters of the impulsive signals, viz. signal-to-noise ratio, crest level, pulse
duration, period and duty cycle. The average value of A falls within ±1.5 dB over
the full range examined for each of the above parameters. However- except for the
decrease in A for repetitive rates < 1 pps (period > 1 sec), no definitive trend in A
with any of the other parameters was evident.
It was mentioned previously that a broadband noise with normally distributed
instantaneous pressures had a crest level of about 10 dB. Thus, any impulsive signal
evaluated with an ISLM must have a crest level greater than 10 dB before it can
produce meaningful test data. This, in turn, implies directly that the accuracy in
reading the meter is limited by the high crest level of the synthetic signals used in the
data acquisition. Manufacturers of impulse sound level meters estimate this inaccuracy
in the meter reading as ±1.0 dB for the highest crest levels employed in this study.
These readings had to be made on the lower part of the ISLM scale. Any objective
correction factor which is not greatly different than this inherent ±1 dB scale reading
error cannot be considered as significant. It is concluded, therefore, that the
A-weighted "SLOW" meter setting can be used to measure directly the A-weighted
equivalent sound level (L ) to within an accuracy of ±1.5 dB for an impulsive signal
with a repetition rate greater than 0.4 pps. However, caution must be exercised on
two factors concerning this conclusion. First, for Impulse Sound Level Meters with a
A-30
-------
conventional (-10 ho +10 dB) meter scale, it is necessary to use the lower portion of
the scale for data acquisition and to use the maximum reading of the sound level meter
for reading fluctuating levels. Secondly, the conclusions do not necessarily apply to
the latest state-of-the-art Sound Level Meters which may employ even more accurate
impulse measuring characteristics and digital readouts or true integration features for
measuring an equivalent level (Lg ) directly. It is anticipated, however, that the
latter type of instruments would, in fact, exhibit even less error when measuring the
true equivalent level (L ) of impulsive sounds in terms of the A-weighted, slow
reading.
Pulse repetition rates lower than 0.4 Hz were not measured in this study. How-
ever, at this pulse rate, the maximum sound level meter reading for each pulse will tend
to approximate the reading obtained on a single isolated pulse with the same character-
144
istics as each of the repetitive pulses. Young and Cohen have shown that for single
cycle sine bursts with burst frequencies greater than 100 Hz (i.e., pulse durations less
than 10 ms), the A-weighted sound exposure level for such a pulse can be obtained
quite accurately by the maximum reading on a sound level meter set to A-weighting,
SLOW. (For lower pulse frequencies, this sound level meter reading will tend to
exceed the true sound exposure level reaching a maximum error of about +8 dB for a
single 20 Hz sine burst.) However, the type of impulsive noise sources of concern for
this study are not expected to involve significant sine pulse components as low as this.
For example, the one-third octave band spectra of the ISO single event impulsive sounds
shown later in Figure B-4 of Appendix B have peak frequencies well above 100 Hz. If
spectral content of an impulse is, in fact, dominant at low frequencies (below 100 Hz),
144
then, according to the results of Young and Cohen the A-weighted sound exposure
level can also be obtained within a maximum error of about 1 dB for pulse frequencies
down to 20 Hz by using the maximum reading on the C-weighting scale. Thus, for the
objective correction factor, an interim recommended procedure is as follows:
A-31
-------
RECOMMENDATION
Until more definite data are available, the objective correction factor for
the measurement of the equivalent (energy average) sound level of impulsive
noise sources shall be assumed equal to zero when the L is based on the
maximum reading on the A-scale (SLOW) of an Impulse Precision Sound
Level Meter. For single isolated pulses, the corresponding equivalent sound
level for N such single events, over a time T (seconds) can be approximated
by
L =- L. _(e) + 10 log N - 10 log T (A-2)
eq Ab
where
L» <-(e) = the energy mean value of the maximum A-weighted
(SLOW) noise level over the N events
When the dominant pulse frequency is below 100 Hz, the C-weighting scale
should be used instead of the A-weighting.
This interim procedure is equivalent to setting the objective correction factor
(the difference between the L of the test signal and the L of the reference signal
eq eq
for the same instrument reading) equal to zero. In any event, a correction factor
would not have been required at all if sound level measurements of transient events
were obtained with a true rms time-integrating meter which measured sound exposure
level.
A-32
-------
APPENDIX B
ISO Round Robin Tests
The most complete set of data on loudness of impulsive noises is provided by
the results of an international cooperative Round Robin test program organized under
the auspices of the International Standards Organization, ISO/TC 43/SC-l, Study
Group B (Secretariat-15) 23, "The Round Robin Test on Evaluation of Loudness Level
of Impulsive Noise." The final report from the organizers, O. Juhl Pedersen, et al,
provides summary data from a portion of the results from over 22 laboratories covering
"close to 400 subjects." More detailed results, from the National Physical Labora-
tory (NPL), included in the summary report, have been reported by Shipton, Evans and
Robinson. Pertinent results from these reports are summarized here.
The test signals employed for these round robin tests consisted of the following
three groups:
Group I: Nine quasi-steady impulsive noise signals recorded from practical
noise sources, e.g., teletype, pneumatic hammer, outboard
motor. Each noise sample has a duration of approximately 1 sec-
ond and is recorded repeatedly alternating with the reference
signal (1/3 octave band of noise at 1 kHz). Intuitively judged,
the noises of this group form a continuum ranging from highly
impulsive to almost steady noises. (Their relative 1/3 octave
band spectra are shown in Figures B-l to.B-3).
Group II: Five noises basically consisting of a single pulse, e.g., from a
gun or a mechanical ram. These noises are recorded as for
Group I with reference signals (1 kHz tone pulses) of approxi-
mately the same duration as the pulse. (Their relative frequency
spectra are shown in Figure B-4).
Group III: Six 1 kHz tone pulses of durations from 5 ms to 160 ms. The
reference signals are 1 kHz tone pulses of durations 10-320 ms.
B-l
-------
Legend
No. * Source
6, Outboard Motor
4, Air Hammer (Silenced)
Mechanical Ram
-60
Frequency in Hertz
Figure B-l. Relative One-Third Octave Band Spectra of ISO Round Robin
Impulsive Noise Samples, Numbers 6, 4, 8 (as Measured By
NPL)66
B-2
-------
No.
Source
1 Card Punch
2 Cement Mill
5 Compressed Air Drill
Oi i i i i' i ill
OQ
~a
-10
O
JB
I
(0
o
t>
o
~a
I
0)
6
Duration =* 1.1 Sec
Frequency in Hertz
-50
-60
Figure B-2. One-Third Octave Band Spectra of ISO Round Robin Impulsive
Noise Samples 1, 2, 5 (As Measured by NPL)
B-3
-------
Legend
No. Source
CO
6
0
-10
-20
-30
-40
-50
-66
3
9
Teletype (No Cover)
Paper Tape Punch
Hamroer on Anvil
Duration =* 1.1 Sec
Frequency
Figure B-3. Relative One-Third Octave Band Spectra of ISO Round Robin Impulsive
Noise Samples 3, 9, 7 (as Measured by NPL)
66
B»4
-------
0
CQ
"O
_v -10
5
0)
E
0)
O
5
.3
-a
o
CQ
-20
-30
6 -40
O
O
I
0)
c
O
-50
-60
-"- - -
_.. ."
--
"--—
- - :
.::.
.. ":
—
--
• -
No.
14
13
10
12
11
f
I
'-'•:
Legend
Source
Simulated Sortie Boom
Gun in Free Field
Electric Typewriter
Explosion Rqm
Hammer on Copper Sheet
..":
j
li
1
I
1
'.
- <
I
'
',
>
t
t
i
'^,
j
W
r IT
1 1
y.
f i
/
/
S
/
^^
k^
/^
/
S
f
•^
X
v^
/
/
/
•j
\
,>
<
/
/
s
'>
•^
/>
te
/
/
Approximate Duration
~ 0.55 Sec
~ 0.05 Sec
~ 0.25 Sec
~ 0.25 Sec
~ 0.6 Sec
\
§•
/
s
~*
\
j
fv
«
^
^ /
V /
s^>
\ f
fc*
\v
V
A
_, \
\
s_
^s
ivir-
/
^
>
\
-
:-
/
*««
j
K
t
-*
^
'-
—
5
i^
^
!!
-
__
*,-
i- -
=
I
1
'
-
: _..;.:IL::.I:
. . — - —
*
: X — —
t
•t -
.-V-~
1^
-i--
'^::=
t--:; -E
h'j"''"'
100
1000
Frequency in Hertz
101
Figure B-4. Relative One-Third Octave Band Spectra of ISO Round Robin
Impulsive Noise Samples 10 to 14 (As Measured by NPL)6t
B-5
-------
The source of the first 14 impulsive sounds is identified on the preceding
Figures. The sounds were presented in repeated A-B sequences at 3 sound levels (55,
75, and 95 dB re 20 |Ll Pa) to the subjects using, in each case, loudspeaker presentations
in presumably free-field or nearly free-field conditions. The subjective data which
will be reported here consist of average values (over subjects) of the difference in
settings of attenuators placed in the test and reference signal channels (i.e,, attenua-
tion of test signal minus attenuation for reference signal) required to achieve equal
loudness between the two signals. This "Equal Loudness Attenuation" for the subject
tests (called ELA , . by Pederson) provided a basic raw measurement of the relative
subjective loudness for each of the test sounds. In order to determine a subjective
correction factor AS from these tests, it was necessary to utilize the additional detailed
data from Shipton, et al, to correct these attenuator settings for the additional
relative difference in the test signals before any relative attenuation was applied.
Thus, as illustrated in Figure B-5, an additional small deviation A accounts for the
difference in L of the reference signal and the test signal before the additional
eq
(ELA , .) attenuation is applied. Thus, as illustrated in the figure below, AS can be
defined by
A =ELA , .
s subj
(B-l)
Reference
Signal
L
eq
dB
Test Signal
LR — ;
A
A.
t
c
J_
t
ELA
subj
-_J__.
LD - L of Reference Signal
K eq °
L = Test Level Before Attenuation
LJ -Test Level Affer Attenuation
(Equal Loudness to I )
R
Figure B-5. Computation of the subjective Correction Factor A From the ISO Data.
77
B-6
-------
The values for A were computed from the detailed data on the reference and
unattenuated test signal levels in Table 2 of Reference 66. It was assumed that these
data apply universally to the ISO average values for ELA , . for the corresponding
sounds. In other words, it was necessary to assume that the relative unattenuated
signal levels from one noise to another were essentially fixed on the Round Robin tapes
and would be reflected in identical variations in each laboratory. This clearly is an
approximation but is not considered unreasonable considering the expected care each
laboratory would take to provide a "flat" reproduction of the (uniform) test tapes pro-
vided by ISO to each laboratory. Table B-l summarizes the reported values of ELA , .
from the ISO report for the nine Group I sounds.
Table B-l
ISO Round Robin Comparisons for ELA , . (dB)
Parameters
Mean
Std. Dev.
Sounds (Group I)
1
12.2
3.5
2
8.9
3.5
3
7.0
4.9
4
7.5
3.9
5
8.4
3.7
6
11.5
3.2
7
8.2
3.3
8
8.7
4.1
9
11.4
3.5
Average
9.31
Std.
Dev.
1.90
The computations for A from the NPL data and the corresponding values for A
are given in Table B-2. The overall grand average of A (including all laboratories in the
ISO figures, all 3 levels, nearly 400 subjects, and for the 9 Group I impulsive noises)
is 12.5 dB. The standard deviation over the 9 average values for each noise is 0.9 dB.
It must be recognized, of course, that this is a highly smoothed statistical result for, as
pointed out in the ISO report, variation in ELA , . values from subject to subject for
any one level and test sound can be 10 to 15 dB. Nevertheless, the central tendency
of the data is clearly indicated by the above values. Considering the necessary assump-
tions required to compute A from these data, it is estimated that the values given in
Table B-l are reliable within better than + 1.0 dB.
B-7
-------
Table B-2
Summary of ComputaMon of A from ISO Round Robin Data
for First Nine (Repetitive) Impulsive Sounds
Row
1
2
3
4
5
6
7
Data Source
I NPL | (66),
V"»
"/>
L (Ref)(d)
eq
L (TesO(e)
eq
\<°
ELA(,ub)(='
..«•>
1 -
79.1
+0.7
(77.1)
75.6
1.5
12.2
13.7
2
77.6
-0.8
(75.6)
73.4
2.2
8.9
11.1
3
77.3
-1.1
(75.3
69.7
5.6
7.0
12.6
Impulsivt
4
77.6
-0.8
(75.6)
69.9
5.7
7.5
13.2
j Sound
5
78.3
-0.1
(76.3)
70.9
5.4
8.4
13.8
6
78.6
+0.2
(76.6)
75.6
1.0
11.5
12.5
7
78.4
0
(76.4)
77.8
3.6
8.2
11.8
8
78.3
-0.1
(76.3)
72.9
3.4
8.7
12.1
9
78.2
-0.2
(76.2)
75.9
0.3
11.4
11.7
Ref.
Signal
(a)
78.4
0
(a)
76.4
•*•
Avg
9.3
,2.5°'
±0.9
(a) Level of Calibration Tone for Reference Signal.
(b) L (Ref) = Impact Sound Level from Table 2b,
AI
Reference 66 (same as dB(Aj)).
(c) A LA] = LAI(Ref) - 78.4 (where 78.4 = LAJ of
Calibration Tone on Reference Channel.
(d) L (Ref) =76.4 +Row 2, Estimated Values of
eq
L for Reference Channels. (76.4 = L of
eq eq
Calibration Signal on Reference Channel.)
(e) From Table 2a, Reference 66 (same as dB(A)
Integrated).
(f) A = L (Ref) - L (Test), Row 3 - Row 4.
r eq eq
(g) From Table 3.3.3.2, Reference 77.
(h) A=ELA . + A , Row 5 + Row 6.'
s subj r
(i) Overall Mean A for all 9 Sounds.
s
B-8
-------
Not considered here is the fact that the values of ELA . reported by Shipton
t . 66 , subj ^
er al, show a variation with presentation level due to the so-called mid-level bulge
in loudness growth.* The effect was relatively small, however, and has been averaged
out in the above figures. Since L data were not available for the 5 single impulsive
eq or
sounds (Group II), A$ values for these sounds could not be established.
Combined Subjective and Objective Corrections From ISO Round Robin Tests
Analogous to the Equal Loudness Attenuation (ELA , .) to achieve subjective
equality of the reference and test signals, there is also an objective Equal "Loudness"
Attenuation (ELA , .) - again adopting Pedersen's terminology - which is the attenuation
of the test signal required to achieve the same response on the objective measuring
instrument as for the reference signal. The comparable objective correction factor A
which we seek will be the difference in L between the test and reference signals to
eq
achieve the same "instrument reading." As with A , there is the same initial difference
in level A between the reference signal and the unattenuated signal and it can be shown
that, for the procedures employed in the 150 tests,
A0=-(ELA +A) , dB (B-2)
obj t
Thus, the quantity ELA , . - ELA , ., reported for the ISO tests, is the same as the
sum of our objective and subjective correction factors (A +A ). This quantity can be
shown to be equal to the equivalent level of the reference signal, when it is adjusted
to the same loudness as the test signal, minus the equivalent level of this same reference
signal, when it is now adjusted to have the same "instrument" reading as the test signal.'
An ideal "instrument" would have a zero value for A + A so that it would
o s
correctly measure the loudness of an impulsive sound. However, a fixed but consistent
"error", represented by a constant non-zero value of A + A could be considered as a
fixed "instrument" error to be corrected out. The critical parameter, therefore, for
At presentation levels of 55, 75, and 95 dB, the average values of ELA ,.
Reference 66 were 9.3, 11.2, and 9.5, respectively. SU J
from
**
When AS + AQ is added to the "instrument" reading of the test signal the resulting
level is the equivalent sound level, L of an equally loud reference signal .
B-9
-------
evaluating the ability of any "instrument" to measure impulse noise, be it an actual
sound level meter (SLM) or a loudness calculation method, would be the standard devia-
tion of the values of A + A about the mean.
o s
Table B-3 summarizes the comparable values of A + A from the ISO data. The
table defines the mean and standard deviation, over noise sources, subjects, and levels
for A + A for each of the ISO data sources and for the variety of objective measure-
o s
ment "instrument" indicated. It appears from these data that A-weighted sound level,
slow (L. _) equivalent sound level (L ), or some form of loudness calculation using,
AS ^ eq
preferably, time-integrated measures of the spectral content, would all have potentially
higher utility and validity than other "instrument"/metric combinations. For the single
event impulsive sounds (Group II), all the measures with the exception of B-weighted
peak impulse or C-weighted peak-hold indicate substantial variation about the mean.
The results of this Round Robin Test can also be compared, in terms of the mean
objective correction factor A with the results from Appendix A. From Table B-2, the
mean sum A + A for A-weighted slow levels is + 11.6. Subtracting the mean sub-
jective correction factor A of 12.5 from Table B-2 gives a mean objective correction
factor A of - 0.9 dB. That is, the average A-weighted slow ISLM reading of the 9
impulsive sounds tested would be 0.9 dB above the average L of these sounds.
eq
This average objective correction factor from the ISO round robin tests of - 0.9 dB
compares well with the average of A of + 0.1 and - 1.4 dB from the two categories of
impulsive signals (repetition rate > 1 pps or 0.4 to 1 pps respectively) reported in
Appendix A,
B-10
-------
Table B-3
Evaluation of ISO Round Robin Data for
Optimum "Instrument" to Evaluate Impulsive Noise' '
"Instrument"
Sound Level
Meter
Ste>
Mk
/ens
. VI
Stevens
Mk. vn
Zwicker
Metric
, (a)
AS
LAF(Q)
LAI (a)
L (a)
eq
LB(PD (C)
LPK
'Lcs
L (b)
eq(C)
L (b)
eq(C)
Lcs
Sounds 1-9
(Group I )
A + A
o s
+ 11.6 (6)
+ 12.1 (8)
+ 11.2 (11)
+ 10.8 (1)
+ 11.9 (1)
+ 11.8 (1)
+ 4.1 (3)
- 0.5 (1)
- 1.5 (1)
- 0.8 (3)
a
dB
1.5
1.8
2.3
1.3
2.4
2.4
2.2
1.4
1.2
2.6
Sounds 10-14
(Group II)
A + A
o s
+ 13.4 (6)
+ 12.5 (7)
+ 8.3 (11)
+ 10.5 '(1)
+ 4.5 (1)
+ 3.6 (1)
+ 1.6 (1)
- 0.2 (1)
+ 8.4 (3)
a
dB
3.6
4.9
3.3
2.6
1.7
1.6
1.8
1.8
7.3
(a)
(b)
(c)
(d)
Maximum peak reading.
One second integration time.
B-weighted peak impulse.
C-weighted peak impulse.
(e)
Number in parentheses signifies
number of laboratories who provided
data for this value.
A + A defines absolute'accuracy
of loudness prediction, cr is standard
deviation about this mean (see text).
B-ll
-------
APPENDIX C
FREQUENCY SPECTRA OF REPEATED TONE BURSTS
Four basic cases for the frequency spectra of transient sounds are illustrated
in Figure C-1. The corresponding Fourier spectra for each of these cases, where the
135
peak amplitude of the pulse is P , can be given as follows:
Single Square Pulse
(C-l)
Repeated Square Pulse ( P(cut)| = P -
max o T
1 +2
sin(nuu T/2)
nou T/2
n =
1/2
(C-2)
Single Tone Burst !P(jtu)l =
sin(ou - DO ) T/2
T/2
(C-3)
Repeated Tone Burst | P(u)t)
where
= p -
max o r
uu
'-I
n = 1
sinfncu - CD ) T/2
0 1
(no) - CD ) T/2
o I
2"
-.1/2
(C-4)
T = Pulse duration
T = Pulse repetition period
u:, = Pulse frequency
= 2TT/T, the pulse repetition frequency
CJD
n = Order of harmonic
C-l
-------
Time History, P(t)
Fourier Spectrg/
P -
o
-T/2 T/2
P _
o
' v
h-l/f
P -
o
T | T/2
2 !
Single Square Pulse
P
o
-T
/2 ! T
r—
/2
— T —
— *-
-t Repeated Square Pulse
/
/
/
\
V
\
\
V~^->^«-^ .
0 2VT
)• Single Tone Burst -^*-
t Repeated Tone Burst
(1).
Figure C-1. Time History and Fourier Spectra of Four Common
Impulsive Wave Forms
C-2
-------
The general shape of the envelope of the frequency spectra is the same in all
cases ~sinx/x. For the single or repeated tone bursts, the spectrum is the same as for
the corresponding case of a single or repeated square pulse but with the peak frequency
shifted to the right to the frequency (uO of the pulsed tone. The 1/2 power bandwidth
.(Af) of the spectrum for the case of the single tone burst can be expressed as Af = lA.
Thus, for a single pulse with only one cycle, T = 2Tr/w and the 1/2 power bandwidth
is equal to the frequency of the pulse itself. Then, a single impulse with only one
cycle will have a very broad spectrum so that its loudness will correspond to the sum-
mation of loudness over many critical bands in the ear. For a repetitive version of such
an impulsive signal, the frequency separation of the sidebands is equal to the pulse
repetition frequency CD, = 2rr/r. The number (N) of harmonics within the same "1/2
power" point on the spectral envelope would be
N =- - = I/duty cycle
C-3
-------
REFERENCES
PART A. ANNOYANCE OF IMPULSIVE OR FLUCTUATING SOUNDS
I. Plutchik, Robert, "A Critical Analysis of the Literature Dealing with the Effect
of Intermittent Sound Stimuli on Performance, Feeling, and Physiology, with
Preliminary Work Toward an Experimental Analysis of the Problem." Report No.
AD 149589, October 1957.
2. Broadbent, D. E. and Robinson, D.W., "Subjective Measurements of the Relative
Annoyance of Simulated Sonic Bangs and Aircraft Noise." J. Sound Vib. 1,
162-174 (1964).
3. Borsky, P.N., "Community Reactions to Sonic Booms in the Oklahoma City Area.
Vol „ Ho Data on Community Reactions and Interpretations,," AMRL-TR-65-37,
Wright Patterson Air Force Base, Ohio, October 1965.
3a. Pearsons, K.S., "Noisiness Judgments of Helicopter Flyovers, " Federal Aviation
Agency Report FAA DS 67-1, January 1967.
4. Shepherd, L.J. and Sutherland, W.W., "Relative Annoyance and Loudness of
Judgment of Various Simulated Sonic Boom Waveforms." NASA CR-1192, 1968.
4a. Rothauser, E.H., Urbanck, G.E. and Pachl, W.P., "Loudness and Annoyance
of Impulsive Office Noise. " IEEE Transactions on Audio and Electroacoustics,
Vol. AU-16, No. 4, 520-522, December (1968).
5. Nixon, C.W., von Gierke, H.E., and Rosinger, G., "Comparative Annoyances
of Approaching Versus Receding Sound Sources." J0 Acoust. Soc. Am. 45, 330 (1969).
6. Robinson, D.W., "The Concept of Noise Pollution Level „" NPL Aero Report Ac 38,
National Physical Laboratory, Teddington, Middlesex, England, March 1969.
7. Kryter, K., "The Effects of Noise on Man." Academic Press, New York, 304, 423,
1970.
8. Keighley, E.C. "Acceptability Criteria for Noise in Large Offices." J. Sound &
Vib. 11, 83-93 (1970).
9. Robinson, D.W., "Towards a Unified System of Noise Assessment. " J. Sound Vib.
14 (3), 279-298(1971).
R-l
-------
REFERENCES (Continued)
PART A. ANNOYANCE OF IMPULSIVE OR FLUCTUATING SOUNDS (Continued)
10. Anderson, C.M.B. and Robinson, D.W., "The Effect of Interruption Rate on the
Annoyance of an Intermittent Noise." NPL Acoustics Report Ac 53, National
Physical Laboratory, Teddington, Middlesex, England, 1971.
11. Eldred, K., "Community Noise." Wyle Laboratories Report No. WR 71-17, U.S.
Environmental Protection Agency Report No. NTID 300.3, 31 December 1971.
lla. International Organization for Standardization, "Assessment of Noise with Respect
to Community Response." ISO Recommendation R 1996, 1971.
lib. International Organization for Standardization, "Assessment of Occupational Noise
Exposure for Hearing Conservation Purposes." ISO Recommendation R 1999, 1971.
lie. Leverton, J.W., "Helicopter Noise - Blade Slap." Part 2, Experimental Results.
National Aeronautics and Space Administration, NASA CR-1983, March 1972.
12. Parry, H.J. and Parry, J.K., "The Interpretation and Meaning of Laboratory
Determinations of the Effect of Duration on the Judged Acceptability of Noise. "
J. Sound and Vib. 20(1), 51-57(1972).
13. Fuchs, G.Lo/ "Integration Time Constant for Annoyance of Impulsive Noises0"
Acustica27, 313-316(1972).
14. Matschat, K0, Muller, E.A. and Zimmerman, G., "Further Developments of
Noise Indices Taking Into Account the Results of the Aircraft Noises Study of
the Deutsche Forschungsgemeinschaft in Munich." Max-Planck-Institut Fur
Stromungsforschung, March 1973,
15. Fuller, H.Co and Robinson, D.W0/ "Subjective Reactions to Steady and Varying
Noise Environments." NPL Acoustics Report AC 62, National Physical Laboratory,
Teddington, Middlesex, England, August 1973.
16. Schomer, Paul D0, "Predicting Community Response to Blast Noise." Construction
Engineering Research Laboratory, Champaign, III., Report No.CERL-TR-E-17,
December 1973.
17. Munch, C.L. and King, R.J., "Communify Acceptance of Helicopter Noise:
Criteria and Application." Sikorsky Aircraft for NASA, NASA CR-132430, (1974)
18. Berry, B.F., Rennie, A0J. and Fuller, H0Co, "Rating Helicopter Noise: the
Feasibility of an "Impulsive Noise Correction" ." National Physical Laboratory
memo for ISO/TC43/SCI/WG2 (Berry and Robinson - 2) 67, October 1975.
R-2
-------
REFERENCES (ConHnued)
PART A. ANNOYANCE OF IMPULSIVE OR FLUCTUATING SOUNDS (ConHnued)
18a. Man-Acoustics & Noise Inc., "Noise Certification Considerations for
Helicopters Based on Laboratory Investigations." Reort for U.S. Department of
Transportation, FAA-RD-76-116, July 1976.
19. Lawton, B.W., "Subjective Assessment of Simulated Helicopter Blade-Slap Noise."
NASA, Langley Research Center, NASA TN D-8359, December 1976.
20. DIN 45645, "Einheitliche Ermittlung des Beurteilungspegels fur Gerauschimmissionen1
(Uniform Calculation of Judged Loundess Level of Noise Emission). (Entwurf 1976.)
21. National Research Council, Committee on Hearing, Bioacoustics and Biomechanics,
Assembly of Behavioral and Social Sciences, "Guidelines for Preparing Environ-
mental Impact Statements on Noise." June 1977.
22. Patterson, J. H., Jr., Mozo, B.T., Schomer, R.D. and Camp, R.T., Jr.,
"Subjective Ratings of Annoyance Produced by Rotary-Wing Aircraft Noise." U.S.
Army Medical Research and Development Command, USAARL Report No. 77-12,
May 1977.
23. Galloway, W.J., "Subjective Response to Simulated and Actual Helicopter Blade
Slap Noise." Bolt Beranek and Newman Report No. 3573 for the National Aero-
nautics and Space Administration, December 1977.
24. Galanter,- E., Popper, R.D. and Perera, T.B., "Annoyance Scales for Simulated
VTOL and CTOL Overflights. " Paper before 94th Meeting, Acoustical Society of
America, Miami, Florida, December 1977.
24a. Galloway, W.J., "Review of the Development of Helicopter Impulsive Assessment
Proposals by ISO TC43/SC1/WG2 - Aircraft Noise." Memorandum Report,
January 1978.
24b. Leverton, J.W., Southwood, B.J. and Pike, A.C., "Rating Helicopter Noise."
Westland Helicopters, Ltd., England, 1978.
24c. Powell, C.A., "A Subjective Field Study of Helicopter Blade-Slap Noise."
National Aeronautics and Space Administration, Langely Research Center, NASA
Technical Memorandum 78758, July 1978.
24d. d'Ambra, F. and Damongeot, A., "Annoyance of Helicopter Impulsive Noise."
S.N.I. Aerospatiale, Marignane, France. Paper from "Helicopter Acoustics, "
NASA Conference Publication 2052, Part II, May 22-24, 1978.
R-3
-------
REFERENCES (Continued)
PART A. ANNOYANCE OF IMPULSIVE OR FLUCTUATING SOUNDS (Continued)
24e. Klump, R. G. and Schmidt, D.R., "Annoyance of Helicopter Bladeslap."
Naval Ocean Systems Center Technical Report 247, July 3, 1978.
24f. Srernfeld, H., Jr., and Doyle, L.B., "Evaluation of the Annoyance Due to
Helicopter Noise." NASA Contractor Report 3001, June 1978.
R-4
-------
REFERENCES (Continued)
PART B. NOISINESS AND LOUDNESS OF IMPULSIVE OR FLUCTUATING SOUNDS
25. Hughes, J.W., "The Threshold of Audition for Short Periods of Stimulation."
Proc. Roy. Soc. (London), B133, 293 (1946).
26. Garner, W.R. and Miller, G.A., "The Masked Threshold of Pure Tones as a
Function of Duration." J. Exp. Psychol. 37, 293-303(1947).
27. Munson, W.A., "The Growth of Auditory Sensation." J. Acoust. Soc. Am. 19,
584-591 (1947).
28. Garner, W. R., "Auditory Thresholds of Short Tones as a Function of Repetition
Rates." J. Acoust. Soc. Am. 19, 600-608 (1947).
29. Garner, W. R., "The Effect of Frequency Spectrum in Temporal Integration of
Energy in the Ear." J. Acoust. Soc. Am. 19, 808-814 (1947).
30. Miller, G.A., "The Perception of Short Bursts of Noise." J. Acoust. Soc. Am.
20,- 513-527 (1948).
31. Gamer, W. R., "The Loudness of Repeated ShortTones." J. Acoust. Soc. Am.
20, 513-527(1948).
32. Garner, W. R., 'The Loudness and Loudness Matching of Short Tones. " J. Acoust.
Soc. Am. 21, 398-401 (1949).
33. Niese, H., "Vorschlag fur die Definition und Messung der Deutlichkeit nach
subjektiven Grundlagen" (Proposal for Definition and Measurement of Distinctive ness
According to a Subjective Basis). Hochfrequenztechn. Elektroakust. 65, 4.
34. Green, D.M., Birdsall, T.G. and Tanner, W.P-, Jr., "Signal Detection as a
Function of Signal Intensity and Duration." J. Acoust. Soc. Am. 29, 523-531,
(1957).
35. Hamilton, P.M., "Noise Masked Thresholds as a Function of Tonal Duration and
Masking Noise Band Width." J. Acoust. Soc. Am. 29, 506-511 (1957).
36. Pollack, I., "Loudness of Periodically Interrupted White Noise." J. Acoust. Soc.
Am. 32:4, 181-185 (1958).
37. Plomp, R. and Bouman, M.A., "Relation Between Hearing Threshold and Duration
for Tone Pulses." J. Acoust. Soc. Am. 31, 749-758 (1959).
R-5
-------
REFERENCES (Confirmed)
PART B. NOISINESS AND LOUDNESS OF IMPULSIVE OR FLUCTUATING SOUNDS
(Continued)
38. Miskolczy-Fodor, R., "Relation Between Loudness and Duration of Tonal Pulses."
II. Response of Normal Ears to Sounds with Noise Sensation." J. Acoust. Soc.
Am. 32:4, 482-486(1960).
39. Niese, H., "Subjektive Messung den Lautstarke von Bandpassrauschen
(Subjective Measurement of Intensity of Bands of Noise). " Hockfrequenztechnik
und Elektroakustik 68 (1960).
40. Zwislocki, J., "Theory of Temporal Auditory Summation." J. Acoust. Soc. Am.
32:8, 1046-1060 (1960).
41. Small, A.M., Brandt, J.F., and Cox, P.G., "Loudness as a Function of Signal
Duration." J. Acoust. Soc. Am. 34, 513-514(1962).
42. Port, E., "Uber die Lautstarke einzelner kurzer Schallimpulse (On the Intensity of
Short Single Sound Impulses)." Beih. Akust., No. 1, 212 (1963).
43. Sheeley, E.C. and Bilger, R.C., "Temporal Integration as a Function of Frequency."
J. Acoust. Soc. Am. 36, 1850-1857(1964),
44. Carter, N.L., "Effect of Repetition Rate on the Loudness of Triangular Transients."
J. Acoust. Soc. Am. 37:2, 308-312 (1965).
45. Zwicker, E., "Temporal Effects in Simultaneous Masking and Loudness." J.
Acoust. Soc. Am. 38, 132-141 (1965).
46. Zepler, E.E. and Harel, J.R.P., "The Loudness of Sonic Booms and Other
Impulsive Sounds." J. Sound Vib. 2, 249-256 (1965).
47. Garrett, R.M., "Determination of the Loudness of Repeated Pulses of Noise."
J. Sound Vib. 2, 42 (1965).
48. Ekman, G., Berglund, B. and Berglund, V., "Loudness as a Function of the
Duration of Auditory Stimulation." Scandinavian Jour, of Psychology 7, 201-208
(1966).
49. Kryter, K.D., "Review of Research and Methods for Measuring Loudness and
Noisiness of Complex Sounds." NASA CR442, 1966.
R-6
-------
REFERENCES (Continued)
PART B. NOISINESS AND LOUDNESS OF IMPULSIVE OR FLUCTUATING SOUNDS
(Continued)
50. Stevens, J .C . and Hall, J . W., "Brightness and Loudness as Functions of Stimulus
Duration." Perception and Psychophysics 1 ,,319-327 (1966).
51. Zwicker, E., "Ein Beitrag zur Lautstarkemessung Impulshaltiger Schalle (A
Contribution to the Measurement of Intensity of Impulsive Sounds)." Acustica 17,
11-22(1966).
52. Pearsons, K.S., "Noisiness Judgments of Helicopter Flyovers. " Report No. FAA
DS-67-1, January 1967.
53. Pearsons, K. and Horonjeff, R., "Category Scaling Judgment Tests on Motor
Vehicle and Aircraft Noise." Report No. FAA DS-67-8, AD 658 755, July 1967.
54. Dubrovskii, N.A. and Tumarkina, L.N., "Investigation of the Human Perception
of Amplitude-Modulated Noise." Soviet Physics-Acoustics 13:1, July-September
1967.
55. Johnson, D.R. and Robinson, D.W., "The Subjective Evaluation of Sonic Bangs."
Acustica 18, 241-258 (1967).
56. Bauer, B.B., Torick, E.L., Rosenheck, A.J. and Allen, R.G., "A Loudness-
Level Monitor for Broadcasting. " IEEE Trans, in Audio and Electroacoustics,
AU-15, 177-182 (1967).
57. Herbert, F., Young, I.M. and Wenner, C.H., "Auditory Flutter Fusion and
Envelope of Signal." J. Acoust. Soc. Am. 44:6, 803-806 (1968).
58. Shepherd, L.J. and Sutherland, W. W., "Relative Annoyance and Loudness of
Judgments of Various Simulated Sonic Boom Waveforms." NASA CR-1192, 1968.
59. Rothauser, E.H., Urbanek, G.E. and Pachl, W. P=, "Loudness and Annoyance of
Impulsive Office Noise." IEEE Transactions in Audio and Electroacoustics, AU-16,
520-522 (1968).
60. Johnson, D.R. and Robinson, D.W., "Propedure for Calculating the Loudness of
Sonic Bangs." Acustica 21, 307-318 (1969).
60a. Shipton, M.S. and Robinson, D.W., "Ambient Noise Limits for Industrial Audi-
ometry." National Physical Laboratory, NPL Acoustics Report Ac 69, April 1969.
R-7
-------
REFERENCES (Continued)
PART B. NOISINESS AND LOUDNESS OF IMPULSIVE OR FLUCTUATING SOUNDS
(Confirmed)
61. Reichardt, W. and Niese, H., "Choice of Sound Duration and Silent Intervals for
Test and Comparison Signals in the Subjective Measurement of Loudness Level."
J. Acoust. Soc. Am. 47:4(2), 1083-1090(1970).
62. Reichardt, W., "Subjective and Objective Measurement of the Loudness Level of
Single and Repeated Impulses." J. Acoust. Soc. Am. 47, 1557-1562 (1970).
63. Fidell, S., Pearsons, K., Grignetti, M. and Green, D., "The Noisiness of
Impulsive Sounds." J. Acoust. Soc. Am. 48:6(1), 1304-1310(1970).
64. Fidell, S. and Pearsons, K.S., " Study of the Audibility of Impulsive Sounds."
NASA CR 1598, 1970. (See also J. Acoust. Soc. Am. 48: 1304-1310, 1970.)
65. Ollerhead, J., "An Evaluation of Methods for Scaling Aircraft Noise Perception."
NASA CR 1883, Wyle Laboratories Report No. WR 70-17, 1971.
66. Shipton, M.S., Evans, D.H. and Robinson, D.W., "An Investigation of the
Loudness of Noises with Impulsive Characteristics." National Physical Laboratory,
NPL Acoustics Report Ac 50, July 1971.
67. Thompson, P.O. and Gales, R. S., "Subjective Judgment of Loudness Level of
Impulsive Noises for the International Round Robin Tests." Paper FF8, Acoustical
Society of America Meeting, October 1971.
68. Zwicker, E. and Fasti, H., "On the Development of the Critical Band." J.
Acoust. Soc. Am. 52:2 (Part 2), 699-702 (1972).
69. Parry, H.J. and Parry, J.K., "The Interpretation and Meaning of Laboratory
Determinations of the Effect of Duration on the Judged Acceptability of Noise."
J. Sound and Vib. 20(1), 51-57(1972),
70. Leverton, J.W., "Helicopter Noise-Blade Slap. Part 2, Experimental Results."
Institute of Sound and Vibration for National Aeronautics and Space Administration,
NASACR-1983, 1972.
71. Fuchs, G.L., "Integration Time Constant for Annoyance of Impulsive Noises."
Acustica 27, 313-3U(L) (1972).
72. Carter, N.L., "Effect of Rise Time and Repetition Rate on the Loudness of Acoustic
Transients." J. Sound and Vib. 21(2), 227-239(1972).
R-8
-------
REFERENCES (Continued)
PART B. NOISINESS AND LOUDNESS OF IMPULSIVE OR FLUCTUATING SOUNDS
(Concluded)
73. Stephens, S.D.G., "Auditory Temporal Integration as a Function of Intensity. "
J. Sound and Vib. 30(1), 109-126(1973).
74. Carter, N.L. and Dunlop, J.I., "The Effects of Rise Time and Repetition Rate on
the Thresholds for Acoustic Transients." J. Sound Vib. 30(3), 359-366 (1973).
75. Boone, M.M., "Loudness Measurements on Pure Tone and Broad Band Impulsive
Sounds." Acustica 29, 198-204(1973).
76. Leverton, John W., "Helicopter Noise - Are Existing Methods Adequate for Rating
Annoyance or Loudness?" J. American Helicopter Society, 41-44, April 1974.
77. Pederson, O. Juhl, Lyregaard, P.E. and Poulsen, T., "The Round Robin Test on
Evaluation of Loudness Level of Impulsive Noise." ISO/TC 43/SC/SG'B1
(Secretariat-15) 23, 1977.
78. Gustafsson, B., "The Loudness of Transient Sounds as a Function of Some Physical
Parameters." J. Sound and Vib. 37, 389-398(1974).
79. Terhardt, E., "On the Perception of Periodic Sound Fluctuations (Roughness)."
Acustica 30, 201-213(1974).
80. Fuller, H.C. and Robinson, D.W., "Temporal Variables in the Assessment of an
Experimental Noise Environment." National Physical Laboratory, NPL Acoustics
Report Ac 72, February 1975.
81. Magliozzi, B., Metzger, F.B., Bausch, W. and King, R.J., "A Comprehensive
Review of Helicopter Noise Literature." Report No. FAA-RD-75-79, June 1975.
82. Isumi, Kiyoto, "Two Experiments on the Perceived Noisiness of Periodically
Intermittent Sounds," Noise Control Engineering 9, 16-23 (1977). Errata for
Fig. 13 and 14 by Personal Communication, October 13, 1978.
R-9
-------
REFERENCES (Continued)
PART C. DETECTION OR PERCEPTION OF IMPULSIVE SOUND
83. Reese, T.S., Kryter, K.D. and Stevens, S.S., "The Relative Annoyance Pro-
duced by Various Bands of Noise." Psycho-AcousHc Lab, Harvard University,
1C-65, 1944.
84. Sherwin, C.W., etal, "Detection of Signals in Noise: A Comparison Between
the Human Detector and an Electronic Detector." J. Acoust. Soc. Am. 28,
617-622 (1956).
85. Marill, T., "Detection Theory and Psychophysics. " MIT Research Lab, Electron
Report No. 319, October 1956.
86. Niese, H., Beitrag zur Relation zwischen Lautstd'rke und Ldstigkeit von Gerauschen
(Contribution to the Relation Between Intensity and Annoyance of Noises)."
Acustica 15, 236-243 (1965).
87. International Organization for Standardization, "Method for Calculating Loudness
Level." ISO/R-532-1966(E).
88. Parnell, J.E., Nagel, D.C. and Parry, H.J., "Growth of Noisiness for Tones and
Bands of Noise at Different Frequencies. " U.S. Department of Transportation,
FAA DS 67-21, 1967.
89. Zwicker, E. and Feldtkeller, R., "Das Ohr als Nachrichtenempfanger, 2 Auflage
(The Ear as a Reception of Messages, 2nd Edition)." S. Hirzel, Stuttgart (Second
Edition), 248, 1967.
90. McGill, W.J., " Variation on Marill's Detection Formula." J. Acoust. Soc. Am.
43, 70-73 (1968).
91. Jeffress, L.A., "Mathematical and Electrical Mode Is of Auditory Detection."
J. Acoust. Soc. Am. 44, 187-203 (1968).
92. Teranishi, R. and Shaw, E.A.G., " External-Ear Acoustic Models with Simple
Geometry." J. Acoust. Soc. Am. 44, 257-263 (1968).
93. Stevens, S.S., "Procedure for Calculating Loudness: Mark VI." J. Acoust. Soc.
Am. 33, 1577-1585 (1961). (See also, "Perceived Level of Noise by Mark VII
and Decibels (E). " J. Acoust. Soc. Am. 51, 575-601 (1972).)
94. Scharf, B., "Loudness" Chapter in Vol. 4, The Handbook of Perceptions, Carterette,
E.C. and Friedman, M.P. (Editors), Academic Press, New York, September 1973.
95. Shaw, E.A.G., Transformation of Sound Pressure Level from the Free Field to the
Eardrum in the Horizontal Plane." J. Acoust. Soc. Am. 56, 1848-1861 (1974).
R-10
-------
REFERENCES (Continued)
PART D. SPEECH INTERFERENCE
96. Miller, G.A. and Licklider, J.C.R., "The Intelligibility of Interrupted Speech."
J. Acoust. Soc. Am. 22, 167-173(1950).
97. Hawley, M.E. and Kryter, K.D., "Effects of Noise on Speech." Chapter 9 in
"Handbook of Noise Control," (C.M. Harris, Ed.), McGraw-Hill, N.Y., 1957.
98. Pollack, I., "Message Procedures for Unfavorable Communication Conditions."
J. Acoust. Soc. Am. 30, 282-285(1958).
99. Williams, C.E. etal., "The Speech Interference Effects of Aircraft Noise." Report
No. FAA DS-67-19, Bolt Beranek and Newman, Inc., 1967.
100. American National Standards Institute, "Methods for the Calculation of the Articulation
Index." ANSI S3.5-1969.
101. Jeffress, L.A., "Masking." In "Foundation of Modern Auditory Theory, Vol. 1."
(J.V. Tobias, Editor), Academic Press, New York, 1970.
102. Scharf, B., "Critical Bands." In "Foundation of Modern Auditory Theory, Vol. 1."
(J.V. Tobias, Editor), Academic Press, New York, 1970.
103. Kryter,- K., "The Effects of Noise on Man." Academic Press, New York, p. 55, 1970.
104. Webster, J.C., "The Effects of Noise on the Hearing of Speech." In: Proceedings of
the International Congress on Noise as a Public Health Problem." Dubrovnik, Yugo-
slavia, Report No. EPA 550/9-73-008, May 13-18, 1973.
105. von Gierke, H. (Task Group Chairman), "Impact Characterization of Noise Including
Implications of Identifying and Achieving Levels of Cumulative Noise Exposure."
U.S. Environmental Protection Agency Aircraft/Airport Noise Study Report, NTID
73.4, 27 July 1973.
106. U.S. Environmental Protection Agency, "Public Health and Welfare Criteria for Noise."
Report No. 550/9-73-002, July 27, 1973.
R-ll
-------
REFERENCES (Continued)
PART E. SLEEP INTERFERENCE
107. Oswald, I., Taylor, A.M., and Triesman, M., "Discriminitive Responses to
Stimulation During Human Sleep." Brain, 83, 440-553 (1960).
108. Schieber, J.P, Mery, J., and Muzet, A., "Etude Analytique en Laboratoire de
I'Influence du Bruit sur le Sommeil." Report of Centre d'Etudes Bioclimatiques du
CNRS, Strasbourg, France, 1968.
109. Lukas, V. S., and Kryter, K.D., "Awakening Effects of Stimulated Sonic Booms and
Subsonic Aircraft Noise." In: "Physiological Effects of Noise." Welch, B.L. and
Welch, A.S. (Editors), Plenum Press, New York, 283-293 (1970).
110. Berry, B. andThiessen, G.J., "Effects of Impulsive Noise on Sleep." National
Research Council of Canada, NRC 11597, 36, 1970.
111. Mery, J ., Muzet, A., and Schieber, J . P., "Effects du Bruit d'Avions sur le Sommeil."
In: Proceedings of 7th International Congress of Acoustics, 3, Budapest, 509-512,
(1971).
(See also numbers 103, 106)
PARTF. HEARING DAMAGE
112. Ward, W. D., Glorig, A., and Sklar, D., "Relation Between Recovery from Temporary
Threshold Shift and Duration of Exposure." J. Acoust. Soc. Am. 31:5, 600-602 (1959).
113. Ward, W. D. , Glorig, A., and Sklar, D., "Temporary Threshold Shift Produced by
Intermittent Exposure to Noise." J. Acoust. Soc. Am. 31:6, 791-794 (1959).
114. Ward, W. D., Selters, W. and Glorig, A., "Exploratory Studies on Temporary
Threshold Shift from Impulses." J. Acoust. Soc. Am. 33:6, 781-793 (1961).
115. Kryter, K., Ward, W.D., Miller, J. D., and Eldredge, D.H., "Hazardous Exposure
to Intermittent and Steady-State Noise." J. Acoust. Soc. Am. 39:3, 451-464 (1966).
116. Fletcher, J.L., "Recovery from Impulse Noise Induced Acoustic Trauma." Report
AD 645 898, U.S. Army Medical Research Laboratory, Ft. Knox, Ky., November 22.
1966.
117. Botsford, J., "Simple Method for Identifying Acceptable Noise Exposures." J. Acoust.
Soc. Am. 42:4, 810-819 (1967).
R-12
-------
REFERENCES (Continued)
PART F. HEARING DAMAGE (Continued)
118. Coles, R0, Garinf-her, G., Hodge, D., and Rice, C., "Criteria for Assessing Hearing
Damage Risk from Impulse-Noise Exposure." U.S. Army Technical Memo 13-67,
Human Engineering Laboratories, Aberdeen Proving Ground, Md., August 1967.
119. Ward, W. D. (Editor), "Proposed Damage-Risk Criterion for Impulse Noise (Gunfire)/1
NAS-NRC CHABA Working Group 57, July 1968.
120. Coles, R., Garinther, G., Lodge, D., and Rice, C., "Hazardous Exposure to Impulse
Noise." J. Acoust. Soc. Am. 43:2, 336-343 (1968).
121. Loeb, M. and Fletcher, J.L., "Impulse Duration and Temporary Threshold Shift."
J. Acoust. Soc. Am. 44:6, 1524-1528(1968).
122. Ward, W.D., "Temporary Threshold Shift and Damage-Risk Criteria for Intermittent
Noise Exposures." J. Acoust. Soc. Am. 48:2(2), 561-574(1970).
123. Luz, G., and Hodge, D.C., "Recovery from Impulse-Noise Induced TTS in Monkeys
and Men: A Descriptive Model." J. Acoust. Soc. Am. 49:6(2), 1770-1777(1971).
124. Cohen, A., Anticaglia, J.R., and Carpenter, P.L., "Temporary Threshold Shift in
Hearing from Exposure to Different Noise Spectra at Equal dBA Level." J. Acoust.
Soc. Am. 51:3(2), 503-507(1972).
125. Burns, W. and Robinson, D.W., "Hearing and Noise in Industry." Her Majesty's
Stationery Office, London (1970).
126. Coles, R. R.A., Rice, C.G., and Martin, A.M., "Noise-Induced Hearing Loss from
Impulse Noise: Present Status." In: Proceedings of the International Congress on Noise
as a Public Health Problem, Dubrovnik, Yugoslavia, May 13-18, 1973. EPA Report
No. 550/9-73-008, 1973.
127. Dieroff, H.G., "Hearing Damage Caused by Very Short, High-Intensity Impulse Noise."
In: Proceedings of the International Congress on Noise as a Public Health Problem,
Dubrovnik, Yugoslavia, May 13-18, 1973. EPA Report No. 550/9-73-008, 1973.
128. Johansson, B., Kylin, B., and Reopstorff, S., "Evaluation of the Hearing Damage
Risk from Intermittent Noise According to the ISO Recommendations." In: Proceedings
of the International Congress on Noise as a Public Health Problem, Dubrovnik,
Yugoslavia, May 13-18, 1973." EPA Report No. 550/9-73-008, 1973.
R-13
-------
REFERENCES (Confirmed)
PARTF. HEARING DAMAGE (Concluded)
129. Passchier-Vermeer, W., "Noise-Induced Hearing Loss from Exposure to Intermittent
and Varying Noise. " In: "Proceedings of the International Congress on Noise as a
Public Health Problem, Dubrovnik, Yugoslavia, May 13-18, 1973. EPA Report No.
550/9-73-008, 1973.
130. U.S. Environmental Protection Agency, "Information on Levels of Environmental Noise
Requisite to Protect Public Health and Welfare with an Adequate Margin of Safety."
Report No. 550/9-74-004, March 1974.
(See also numbers 103, 105, 106)
PART G. MEASUREMENT OF IMPULSIVE SOUNDS
131. Beranek, L. L., "Acoustic Measurements. " Chapters lOand 11. John Wiley
&Sons, Inc. New York (1949).
132. Beranek, L. L. (Ed.), "Noise Reduction. " Chapters 5, 6, and 7. McGraw-
Hill Book Company, Inc. New York (1960).
133. Thiessen, G. J. and Subbarao, K., 'Effect of Reverberation on Reaction to
Impact Noise. " Fourth International Congress on Acoustics, Copenhagen,
21-28 August 1962.
134. International Electrotechnical Commission, 'Precision Sound Level Meters. "
Publication 179, 1965.
135. Crocker, M. and Sutherland, L. C., "Instrumentation Requirements for
Measurement of Sonic Boom and Blast Waves - A Theoretical Study. "
J. Sound Vib. 7:3 (1968).
136. Hewlett Packard, "Acoustics Handbook Application Note 100. "
November 1968.
137. Bruel & Kjaer, "Instructions and Applications - Impulse Precision Sound Level
Meter Type 2204/S"Jun'? (1970).
138. Peterson, A. and Gross, E., "Handbook of Noise Measurement. " General
Radio, 1972.
139. Hempstock, T. I., Powell, J. A., and Else, D., "A Note on the
Limitations in the Use of the Impulse Precision Sound Level Meter. " Applied
Acoustics (5), 141-144 (1972).
R-14
-------
REFERENCES (Concluded)
PART G. MEASUREMENT OF IMPULSIVE SOUNDS (Concluded)
140. Olesen, H.P. and Zaveri, K., "Measurements of Averaging Times of Level
Recorders Types 2305 and 2307. " B & K Instruments Technical Review, No. 1,
1974.
141. Kundert, W.R., "The Impulse Sound-Level Meter - What's It All About?"
Sound and Vibration, 50-53, March 1974.
142. Wahrmann, C.G. and Broch, J.T., "On the Averaging Time of RMS Measure-
ments." B& K Instruments Technical Review, No. 2, 1975.
143. Zaveri, K., "Averaging Time of Level Recorder Type 2306 and 'Fast1 and 'Slow'
Response of Level Recorders 2305/06/07." B & K Instruments Technical Review,
No. 2, 1975.
144. Young, R.W. and Cowen, S.J., "Responses of Sound Level Meters to Impulsive
Sounds." Paper before 91st Meeting, Acoustical Society of America, Washington,
D.C., April 1976.
145. Martin, R., "The Impulse Sound Level Meter and Proposals for Its Use in Germany. "
Paper before Inter-Noise 76, Washington, D.C., 5-7 April 1976.
146. Bruel, Per V., "Noise, Do We Measure It Correctly?" Noise Control Engineering
8, 52-60(1977).
147. Schomer, P., "Evaluation of C-Weighted L for Assessment of Impulse Noise. "
J. Acoust. Soc. Am. 62, 396-399 (1977).
148. Reference for USASI Noise - General Radio, "Type 1382, Random-Noise Generator
20 Hz-50 kHz, "April 1968.
R-15
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
. REPORT NO.
EPA 550/9-79-103
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Annoyance, Loudness, and Measurement of Repetitive
Type Impulsive Noise Sources
5. REPORT DATE
November 1979
6. PERFORMING ORGANIZATION CODE
7. AUTHOfl(S)
L.C. Sutherland, R.E. Burke
^. PERFORMING ORGANIZATION REPORT NO
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Wyle Research
El Segundo, California 90245
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-01-4694
12. SPONSORING AGENCY NAME AND ADDRESS
13. TYPE OF REPORT AND PERIOD COVERED
U.S. Environmental Protection Agency
Office of Noise Abatement and Control (ANR-471)
Washington, D.C. 20460
TY.PE O
Final
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This study was undertaken to evaluate subjective and objective aspects of
moderate levels of noise from impulsive sources. The study excluded evaluation of
hearing damage risk or annoyance from building vibration by high level impulsive
noise, which were covered by recent recommendations of the National Research Council,
Committee on Hearing Bioacoustics and Biomechanics, Working Group 69. While the
study included original investigations into some of the objective aspects of im-
pulsive noise, a detailed review of the literature on the subjective aspects was
emphasized . Based on this available literature, the annoyance and loudness from a
wide variety of repetitive impulse noises were evaluated. These results were
applied to the evaluation of impulsive noise from a number of specific noise sources.
Based on the most pertinent literature, it is tentatively concluded that a subjective
impulse correction factor of +7 dB applied to the A-weighted equivalent sound levels
of these types of repetitive impulsive noise sources would better define their effec-
tive level in terms of annoyance reactions. No additional correction is identified
at this time for crest level or repetition rate. Research on subjective correction
factors for helicopter blade slap is also reviewed and potential reasons for the
smaller subjective correction factors (i.e., 0 to 6 dB) for annoyance response to
(CONTINUED ON BACK OF PAGE)
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
COS AT I Field/Group
Noise, Annoyance, Impulse Noise, Loudness,
Calculation Methods
8. DISTRIBUTION STATEMENT
Limited supply available at ANR-471,
Washington, D.C. 20460 EPA-ONAC
at. NTIS
19. SECURITY CLASS (This Report}
Unclassified
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
GOVERNMENT PRINTING OFFICE 1979 -311-132/150
-------
this type of sound are discussed. It is recommended that refinements to this subjective
correction factor be based on the use of standard loudness calculation methods (Stevens
Mark VII or Zwicker) modified to include provision for a shorter time constant to reflect
subjective response to short duration impulsive sounds.
The study also included a brief experimental evaluation of the measurement of a
wide variety of simulated repetitive impulsive-type signals varying in duty cycle,
repetition rate, pulse frequency, and ratio of peak impulse signal level to continuous
background noise level. When repetitive impulses are measured using maximum values of
A-weighted (slow) readings on an Impulse Sound Level Meter, no objective correction is
necessary in order to measure, with an accuracy of +1.5 dB, the equivalent level (Leq)
of the wide variety of impulsive signals investigated.
------- |