Noise is typically defined as sound that is unwanted, as perceived by a listener. Lovers of motorcycles may not consider the revving of a Harley to be noise, but rather music to their ears. Unwanted sound almost always conveys some information: a newcomer to Holland may not perceive the babble of conversation in neighboring offices as noise until he/she begins learning enough Dutch to be distracted. For this reason, it is possible to "quiet" a space by actually adding sound in the form of background noise, such as the whir of a fan or the hiss of a loudspeaker broadcasting white noise, because the information-carrying content of the sound is reduced. Moreover, individual sensitivities to certain types of noise may be related to listeners' ear physiology (e.g., "better ears") or psychology (e.g., automotive "backfires" perceived by a war survivor). Adaptation over time allows city dwellers to ignore traffic din while conversely being forced into sleeplessness by rural crickets' hum.
Regardless of the human perceptual component, all noise begins as oscillations of air molecules: an alternating squeezing and expanding of the air that gives rise to a propagating acoustic wave. To create a sound, a vibrating object must first cause the air in contact with it to undergo rapid compressions and decompressions; a buzzing object in a bell jar evacuated of air makes no sound because there is little air around it to be set in motion. Thus quieting noisy objects at the source by reducing either their vibrational amplitude or their ability to couple into the air is a common noise-control strategy. Since sounds must propagate through a medium -- such as air, water, or solid wall -- methods to interrupt the propagation path may aid noise suppression. And the receiver's environment can play a large part in noise reception, too. A listener seated in an echoey room will hear more noise from the same sound source than will a listener in a plush, curtained room, because of multiple reflections of the sound from the walls.
Noise fluctuates in amplitude and in frequency content as a function of time. Methods of quantifying noise have thus been developed to take these fluctuations into account, as well as the associated human response to them. Noise regulations and standards have likewise arisen based upon these quantification methods.
The two principle components of noise may be thought of as the "loud-soft" axis and the "high-low" axis, and although they are coupled together in human perception to some extent, they can be separately examined. The "softest," or quietest sound that humans can perceive is nearly the same as the softest sound that all animals (or insects) can perceive: the vibrational amplitude of the air at zero decibels (the softest sound) is only about the diameter of a hydrogen atom, and any further sensitivity of the listener would be overcome by the ambient "hiss" of the Brownian motion associated with the thermal agitation of molecules. Actually, zero decibels is defined as the average softest sound perceived by the tens of thousands of WWII-era soldiers whose hearing tests formed the basis for many acoustic standards.
Since sound intensity is measured, like earthquakes, on a logarithmic scale, each increment of ten decibels amounts to a multiplication by ten of the energy present in the sound. Casual conversation usually takes place around 60 decibels, a moderately loud sound (e.g., forte on a trombone) around 100 decibels, a very loud sound (e.g., Grateful Dead concert) at 110 decibels, and the threshold of pain around 120 decibels. Thus the energy in a 120 decibel sound is 1,000,000,000,000 times that of the softest sound (10^12). Incidentally, decibels are one tenth of a bel (named after Alexander Graham Bell), and so the "B" is capitalized in the abbreviation for decibels, dB.
On the "high-low" axis, it is important to recognize that human hearing is not uniform as a function of frequency. Our ears are relatively insensitive to low tones, and to very high ones, with a broad maximum in sensitivity in the range of audio frequencies corresponding to consonants. The oft-quoted range of hearing is 20 to 20,000 Hz (one hertz, abbreviated Hz, is one vibration per second, named after Heinrich Hertz). Below 20 Hz, the predominant sensation is one of vibration, not auditory sound. On the high end, newborn babies can typically hear 25,000 Hz and above (and often cry at the ultrasonic motion detectors in stores which are unheard by their mystified parents), but by college age that maximum has fallen to 18,000 Hz and by retirement age, to below 10,000 Hz (n.b.: The process of progressive high-frequency hearing loss is a natural one, called presbycusis, that is independent of exposure to noise. It is not just nostalgia; the music really did sound sweeter in your youth!). Of course, we know dog whistles and deer sirens are audible to certain animals but are silent for humans.
Figure 1: The frequency weighting curve for the "A" settings on sound level meters. This setting allows the meter to simulate the properties of the human ear.
Figure 1 shows the curve, developed from the same WWII-era soldiers, that is used to adjust the sensitivity of microphones to simulate the frequency response of the human ear. Sound level meters are devices that quantify noise, and their "A-weighting" setting bumps up or down, by the number of decibels shown on the vertical axis of Figure 1, any sounds with the frequencies shown on the horizontal axis. Thus a 100 Hz tone would have to be "cranked up" by about 20 dB to sound as loud as a 1000 Hz tone does. Some stereo systems do this automatically with a "bass boost" feature, and concert hall designers work hard to ensure that low frequencies produced on stage reach the audience with more power than the mid-frequency sounds. Incidentally, the broad maximum in sensitivity between 1000 and 4000 Hz is produced by the "organ pipe" resonance of the human ear canal, and human speech exploits this anatomical feature for information-carrying purposes.
Given that sound level meters can register the "loud-soft" axis of noise in decibels (dB), and compensate for the frequency characteristics of human ears by A-weighting (dBA), how is the fact of noise fluctuation as a function of time accounted for? Figure 2 shows A-weighted decibels (dBA) of the noise of a trash truck moving down Elm Street in Swarthmore, PA, at around 8:00 am, Friday, June 1, 2001. The absolute maximum noise level, which may last only a small fraction of a second, is one indicator of the noise content, but this measure would not differentiate between noise spikes that repeat and those which occur rarely. The minimum noise level might be considered the background level in the absence of truck noise, except that it is determined by distant traffic sounds and nearby insect noises and hence fluctuates significantly, even without the trucks.
Figure 2: Sound level meter readings (dBA) as a function of time for a trash truck on Elm Street passing by Dan West's driveway. Click here to listen to the sound (2.6 MB .wav file).
The average noise level, or, equivalently, the dBA level of a constant hiss that would produce the same overall amount of noise as the fluctuating sounds did, is called Leq, the "equivalent noise level." Professional-grade sound level meters calculate this quantity over a time interval starting from when a button on the meter is pressed. The noise level that is exceeded 50% the time during that measuring interval is the median level, called L50. The mean (Leq) and median (L50) levels are the same for noises fluctuating symmetrically in time, but this is rarely the case for real noises. The difference between the mean level, Leq, and the median level, L50, can indicate how the statistics of the noise are skewed toward the softer or louder end of the range. But neither of these descriptors will properly account for the effect of the loudest (and softest) sounds: like wild temperature fluctuations, an average value says little about whether the levels encountered were tolerable or intolerable.
Figure 3: Histogram of trash truck sounds from Figure 2, with various descriptors shown. Leq is the average noise level, and Ln is the noise level exceeded n% of the time (during the measurement interval).
Figure 3 shows the complete histogram of the sounds in 10-dBA bins from softest to loudest for the trash truck noise in Figure 2. In conformance with the concept of L50 (the noise level exceeded 50% of the time), it is possible to calculate Ln, the noise level exceeded n% of the time. L1 is nearly the loudest noise recorded during a particular measurement period, since it is the level exceeded only 1% of the time; L99 is nearly the least loud sound, since it is exceed 99% of the time, and so on. While a complete statistical description of the noise would provide the most information, it would be ungainly, so often only the following descriptors are calculated: L90, L50, L10, and Leq. With these four values, the range of noise levels is specified adequately for most purposes.
So now we come to noise regulations and the community response to noise. Various municipalities, government organizations, and international standards makers have tried to define the links between the physical measurements of noise and peoples' reactions to it. One common thread is that comfortable speech levels fall in the range 60-65 dBA, so noise levels that frequently rise into this range reduce speech intelligibility. Thus many regulations specify average noise values Leq of less than 65 dBA. Another thread is that the same noise level that may be tolerable in the daytime is considered intolerable at night, when one is trying to sleep. Therefore many regulations add a penalty of 10 dBA to the measured noise values if they occur between 11 pm and 7 am. Likewise, impulsive noises such as bangs that can provoke a "startle effect" are increased 13 dBA over their measured values (studies show that this amount matches the annoyance of traffic noise), on top of any nighttime penalty. If the time over which the average noise level is calculated is 24 hours, and these penalties are added depending upon the time of day or impulsive nature of individual noise events, then the resulting weighted average is called Ldn (a day-night weighted average noise level, in dBA).
Figure 4: Plot of percent of community residents highly annoyed by sounds with different average noise levels.
Figure 4 shows data, taken over many studies with many subjects, that indicate what percentage of a population judges noise to be too loud, based upon the statistical descriptor Ldn (in dBA). Due to the differences in individuals' sensitivity to noise, there is a continuous increase in the percentage judging a noise as too loud that rises steeply with increasing noise level. In practice, there will always be some people who find ANY noise too loud, and others who won't be bothered even by very loud noises.
|
Authority |
Noise Standard |
|
EPA (1980) |
Ldn less than or equal to 55 dBA outdoors Ldn < or = 45 dBA indoors |
|
HUD |
for building interior, with windows open: Leq < or = 55 dBA for more than 60 minutes in any 24 hour period Leq < or = 45 dBA for more than 30 minutes from 11 pm to 7 am Leq < or = 45 dBA for more than 8 hours in any 24 hour period |
|
FICON (Fed. Interagency Committee on Urban Noise) |
Ldn < or = 65 dBA residential |
|
World Health Organization |
Leq < or = 55 dBA during daylight hours |
|
US Dept. of Transportation |
L10 < or = 70 dBA residences and schools |
|
US Air Force |
Leq < or = 65-70 dBA |
|
City of Chester, PA |
Max level (residential): 55 dBA Max level (commercial): 62 dBA |
|
Ridley Township, PA |
"At no point on the boundary ... of a commercial facility ... shall the sound pressure level ... exceed the maximum permitted sound levels: Frequency (Hz) dBA 36.5 76 63 74 125 68 250 63 500 57 1000 52 2000 45 4000 38 8000 32 |
|
Gainesville, FL |
Maximum permissible noise level is 61 dBA (day) or 55 dBA (night) |
Table 1: A sampling of noise regulations and criteria from a variety of sources.
Table 1 shows various noise ordinances based upon the descriptors described above, as well as some others not mentioned (for example, regulations that single out noise in a particular range of frequencies and specify that it must not exceed some value). For the Science Center construction project at Swarthmore College, there are no local noise ordinances that apply. Federal ordinances are not applicable since the College is private property, and international standards are for guidance only. Nevertheless, noise monitoring will produce data whose statistics can be compared against the various noise ordinances of Table 1.
As mentioned previously, acoustics consultants employ solutions to the source, propagation path, and receiver (listener) locations in an attempt to reduce the noise. On the source side, fixing a broken muffler, asking a driver to slow down before hitting the speed bump, and in general keeping track of the loudest sources may be effective. For the noise propagation path, different techniques are required. For example, barriers (such as the walls along highways such as the Blue Route) cast quieter "acoustic shadows" behind them. Since sounds expand spherically outward from a noise source, every doubling of distance between source and receiver quadruples the surface area of the sphere and reduces the noise level at the listener by at least 6 dB. Avoiding acoustically "leaky" paths into the house helps, since a calculation will show that as much sound will pour through a one-square-inch hole as from a 10-square-foot wall. Options for treating the receiving (listening) location might range from ear plugs (not appropriate for residents, but certainly appropriate for jackhammer operators) to exotic technologies such as anti-sound. Reducing the reverberance of interior spaces by adding acoustically absorbing materials inside listener's rooms, and providing constant background noise (e.g. fan) to mask individual noise events, are also time-tested methods.
Every individual's noise sensitivity is different, and every noise source, propagation path, and interior space is in a unique acoustical arrangement. Thus no list of possible solutions can be complete, and certainly one technique cannot be equally successful in all situations. Only an experienced acoustical consultant, one who can tailor the solution to meet the individual needs of a client, is likely to find a satisfactory solution.
Comments on (including suggestions for improvement of) this document will be gratefully received by the author.
© 2001 by Erich Carr Everbach
Additional wonderful information about noise is available from http://www.nonoise.org/
Send message to the chair of the Science Project User's Group , Rachel Merz (rmerz1@swarthmore.edu)
last updated 7/26/01
webmaster