Generally speaking, noise consists of sound at many different frequencies right across the entire audible spectrum. As the human ear is more sensitive to certain frequencies than others, the level of disturbance is dependant on the particular spectral content of the noise. We are going to look at several different ways of objectively determining just how noisy a sound is perceived to be. A significant amount of work has been done in this area, particularly in the early seventies, and there are a number of accepted techniques in use.

The human ear is most sensitive to sounds in the 500Hz to 4000Hz frequency range and less so for sounds above and below those frequencies. This area of sensitivity corresponds to the human speech band. This non-uniformity in the ears response means that the threshold of audibility for sounds of different frequencies will vary. Thus, by referencing a particular sound level, SPL and SIL values do not consider the ears frequency response at all. In order to take this into consideration, a further modification of the SPL or SIL value is required.

Equal Loudness Contours

Suppose we started with a 1000Hz tone at the threshold of audibility, and used this as our reference. The threshold of other frequencies can then be determined and plotted on a graph. If we increased the 1000Hz tone to, say, 40dB, other frequencies could then be adjusted until they were judged equally as loud. Thus a set of Equal Loudness Contours could be built up, becoming a nomogram defining a new scale, the loudness level, whose units are the phon.

Figure 1 - Equal loudness curves showing how the ear's response differs with frequency.

The phon is a true measure of the response of the human ear. At 1000Hz the dB and phon values are identical. If the sound level of a different frequency is known, its loudness level can be interpolated from this graph. Thus we can now directly compare sounds at different frequencies.

NOTE: You can see from the diagram that it requires over a million times more power to produce an audible sound at 30Hz than at 3000Hz.

The Loudness Scale and the Sone =

As the apparent loudness of a sound is not directly proportional to the sounds loudness level (a doubling of subjective loudness results in an average increase of about 6 phons), subjective experiments have been performed in order to establish a scale on which a doubling of the number of loudness units doubles the subjective loudness, a trebling of loudness units trebles the subjective loudness, and so on.

The unit of this scale is the sone. It is defined as the loudness of a 1000Hz tone having an intensity of 40dB. Thus the sone is equal to the loudness of any sound having a loudness of 40 phons. The mathematical relationship between the two is given as;

S = 2 (P - 40)/10

or

log(S) = 0.0301 (P - 40)

Where:
P represents the loudness level in phons, and
S the loudness in sones.

How is Noise Measured?

Basically, there are two different instruments to measure noise exposures: the sound level meter and the dosimeter. A sound level meter is a device that measures the intensity of sound at a given moment. Since sound level meters provide a measure of sound intensity at only one point in time, it is generally necessary to take a number of measurements at different times during the day to estimate noise exposure over a workday. If noise levels fluctuate, the amount of time noise remains at each of the various measured levels must be determined.

Figure 2 - An example sound level meter.

To estimate employee noise exposures with a sound level meter it is also generally necessary to take several measurements at different locations within the workplace. After appropriate sound level meter readings are obtained, people sometimes draw "maps" of the sound levels within different areas of the workplace. By using a sound level "map" and information on employee locations throughout the day, estimates of individual exposure levels can be developed. This measurement method is generally referred to as "area" noise monitoring.

A dosimeter is like a sound level meter except that it stores sound level measurements and integrates these measurements over time, providing an average noise exposure reading for a given period of time, such as an 8-hour workday. With a dosimeter, a microphone is attached to the employee's clothing and the exposure measurement is simply read at the end of the desired time period. A reader may be used to read-out the dosimeter's measurements. Since the dosimeter is worn by the employee, it measures noise levels in those locations in which the employee travels. A sound level meter can also be positioned within the immediate vicinity of the exposed worker to obtain an individual exposure estimate. Such procedures are generally referred to as "personal" noise monitoring.

Area monitoring can be used to estimate noise exposure when the noise levels are relatively constant and employees are not mobile. In workplaces where employees move about in different areas or where the noise intensity tends to fluctuate over time, noise exposure is generally more accurately estimated by the personal monitoring approach.

In situations where personal monitoring is appropriate, proper positioning of the microphone is necessary to obtain accurate measurements. With a dosimeter, the microphone is generally located on the shoulder and remains in that position for the entire workday. With a sound level meter, the microphone is stationed near the employee's head, and the instrument is usually held by an individual who follows the employee as he or she moves about.

Up until this point, we have been concerned with comparing sounds of only one single frequency, pure sounds. Most sounds however consist of many interacting frequencies, and are called complex sounds. Even musical notes consist of a fundamental frequency, which give it a pitch, and a great number of harmonics (a harmonic being an integer multiple of the fundamental). It is the relative strength of these harmonics that give a sound its timbre.

Sound Weighting Curves

In order to compare different complex sounds, we have to measure their entire spectrum. Most often measurements are taken for each octave or 1/3 octave, starting at 125Hz or sometimes as low as 63Hz.

Using the phon, it is possible to calculate a single figure that refers to the loudness of a complex sound. This can be done by deriving a weighting for the level measured in each band (using the phon scale) and taking the average to produce a result.

As discussed earlier, the ear's response varies with both frequency and level. Thus a system of sound weighting curves are used.

Figure 3 - Sound weighting curves used to derive a single value from a complex spectrum of values based on the ear's response at different frequencies.

• A weighting - used for loudness levels below 55 phons (dBA)
• B weighting - used for loudness levels between 55 and 85 phons (dBB)
• C weighting - used for loudness levels above 85 phons (dBC)
• D weighting - used to account for the increase in annoyance produced by the high frequency whine present in the noise produced by modern aircraft frequencies (dBD)

More recent work has not substantiated these historical associations so that these weightings are now largely conventional. Furthermore, the A-weighting (dBA) seems no longer restricted to low level sounds as it is frequently specified for rating sounds irrespective of level.

An E weighting has also been proposed (Stevens, 1972) using a reference tone of 3150Hz, a much more sensitive frequency region of the ear, however, this is not in common use. Of interest, German standards use DIN-phon curves; three curves based on the response at 60-130dB, 30-60dB and below 30dB at 1000Hz.

NOTE: It is important to remember that you can't simply add or subtract values given in decibels unless that value is given as a decibel reduction. You must first convert them back to their absolute measures, add those and then convert back to decibels. This is shown in the following formula:

SILAve = 10 * log(S(10^SIL/10)

With regard to sound weightings, however, they are given specifically as a decibel reduction. Therefore they can be simply subtracted from the measured sound levels and the results averaged using the method described above.

Noise Rating Curves

In a manner similar to the dBA, if the spectrum of a broad-band noise source is known, the degree of its disturbing effect can be expressed as a single figure, a Noise Rating (NR) number [Fig 4]. This is an internationally accepted standard and is given by the curve just touching the highest point on the noise's spectrum. The slope of the NR curves are based on Equal Loudness Contours but, whilst the latter is used for pure tones, the NR curves are used for rating broad-band noises.

Figure 4 - Noise Rating (NR) curves.

Related Links

• Selecting a Sound Level Meter
  

http://www.noisenet.co.uk/Noise_Instrumentation.htm

• World Health Organisation Media Release on Occupational and Community Noise
  

http://www.who.int/mediacentre/factsheets/fs258/en/