The ‘A’ Frequency Weighting
Ken Scannell Non refereed Noise and Sound Services, Spectrum House, 1 Elegans Avenue, St Ives NSW 2075 Previously published as part of an Acoustics Forum in Acoustics Australia, Vol 31 No 1, April 2003, aimed at promoting discussion. Reprinted with permission.
Introduction Generally, the low frequency bass Figure 1, this approximately follows tones (i.e. 50 to 250 Hz) sound the inverted Fletcher and Munson The ‘A’ frequency weighting is used slightly quieter than the tones in 40 phon curve (± 3 dB). The 40 extensively in many acoustical the mid audio frequency range (i.e. phon curve is based on the noise measurements. Although 1 to 4 kHz). Experiments were subjectively reported equal almost exclusively used, it is often carried out by Harvey Fletcher [1] loudness magnitudes at various misunderstood or incorrectly at the Bell Telephone Laboratories frequencies relative to 40 dB at 1 defined even by those who would in New York, in the early 1930s to kHz. be expected to have a better determine how loud tones of knowledge. It is commonly stated different frequencies sounded The symbol for the ‘A’ frequency in glossaries, even in official subjectively. A series of curves on a weighted sound pressure level,
documents or textbooks on graph were drawn from these measured in decibels is ‘L PA ’ [3] acoustics or noise, as “a scale that experimental results. These become although the common abbreviation simulates the response of the flatter in frequency with higher is dBA or dB(A). Either of the two abbreviations could be used but the symbol is preferred as this places the ‘A’ with the level and not with the decibel, which incorrectly implies there are different types of decibels.
Limitations of ‘A’ weighting Due to its simplicity and convenience, the ‘A’ frequency weighting has become popular and it is an often used frequency weighting for many different noise sources. It is used for all types of noise assessments from Figure 1. The ‘A’ Weighted frequency filter relative to occupational noise, building 0 dB at 1 kHz. acoustics, loudness assessments and noise annoyance assessments. human ear” or similar erroneous sound pressure levels and are The World Health Organization nonsense. known as equal loudness contours. (WHO) [4] has recognised that the From these contours, three curves ‘A’ frequency weighting is an known as A, B, and C frequency overall value which may simulate Origins of the A weightings were developed for use neither the spectral selectivity of weighting curve in sound level meters. These human hearing nor its non linear frequency weightings were specified relation to sound intensity. Quite The human hearing system is not in an American Standard for wrong and totally misleading as sensitive to all sounds if they sound level meters in 1936 [2]. The statements in glossaries are vary in pitch or frequency. ‘A’ frequency weighting is shown in
NEW ZEALAND ACOUSTICS Vol. 16, No. 2 19 commonly given for the ‘A’ normal ear were desired”. first impressions but can vary by up frequency weighting such as “The to 11 dB in the low frequency (e.g. ‘A’ frequency weighting adjusts the The equal loudness contours were 100 Hz) range. noise level to the subjective re determined under more response of the Even if the ‘A’ human ear” or frequency reference is made weighting could to ‘A weighted be used as a good decibels’, which, universal of course do not predictor of exist and should loudness it is not be expressed as ‘A’ a good predictor frequency of noise weighted sound annoyance, levels in decibels. particularly for sounds which Fletcher and differ from those Munson derived which are the original equal medium level, loudness curves broadband mid using only eleven audio frequency, observers who and have constant listened to pure Figure 2. The ‘A’ Frequency Weighting and the Equal temporal tones through Loudness Contours from Fletcher and Munson and characteristics. headphones. In their paper Robinson and Dadson. It is often stated Fletcher and Munson (1933) stated stringent conditions in 1955 using that the ‘A’ frequency weighting “…it would be necessary to increase ninety subjects. The re determined follows the 40 phon equal loudness the size of the group if values more equal loudness contour curves are contour. The confusion comes representative of the average similar to the original curves on from the fact that there are two sets
NOISE CONTROL SERVICES ACOUSTIC PRODUCTS AND DESIGN
• Products for industrial and commercial noise control • Acoustic design • Turn key acoustic solutions
PO Box 82 126, Highland Park 6 Canon Place, Pakuranga AUCKLAND Phone: +64 9 576 4319 Fax: +64 9 576 3806 E Mail: [email protected]
20 Vol. 16, No. 2 NEW ZEALAND ACOUSTICS
of equal loudness contours – one level can be realised by considering risk of occupational hearing from Fletcher and Munson and the simple case of a ‘dripping tap’ damage and the A’ frequency another from Robinson and noise when trying to sleep. weighted exposure to noise. It is Dadson (1956) [5]. The ‘A’ however not a good frequency weighting frequency filter is close The ‘A’ frequency weighting weighting to use when assessing to the Fletcher and Munson 40 should be used for occupational annoyance from noise which is phon curve but varies by up to 8 noise assessments (except peak predominantly low frequency (i.e. dB at low frequencies from the noise assessments) because there below about 250 Hz). ‘more representative’ Robinson are ‘known’ relationships between and Dadson 40 phon curve. This is the statistical risks of hearing a significant difference as it damage and the overall long term References represents close to a 50% change in ‘A’ frequency weighted noise the perception of subjective exposure level [7]. H. Fletcher and W.A. Munson , loudness. The two 40 phon curves, “Loudness: it’s definition, at the low frequency end of the The ‘A’ frequency weighting has, measurement and calculation”, J. spectrum are compared to the ‘A’ unfortunately, never been changed Acoust. Soc. Am. 5, 82 105 (1933). frequency weighting in Figure 2. from the 1936 American Standard even though it was based on results American Standard ASA Many noise sources in the where Fletcher and Munson Z24.2. American National environment are low frequency. indicated that they were not Standard for Sound Level Meters When assessing these noise sources necessarily representative of the (1936). the ‘A’ weighting frequency filter average normal ear. This was later can be regarded as a high pass filter proved to be the case by Robinson Australian Standard, Acoustics with a cut off frequency (10 dB and Dadson. Hence the ‘A’ – Description and down point) at about 250 Hz. frequency weighting is not even a Hence, where a noise source is rough approximation (i.e. about Measurement of dominated by low frequency, the 50% error) to the response of the Environmental Noise Part 1: use of the ‘A’ frequency weighting human ear at 40 phon. General Procedures (1997). gives a poor indication of loudness and an abysmal indication of noise B. Berglund, T. Lindvall & D. annoyance. Summary and H. Schwela (Eds) Guidelines for Community Noise World Health Annoyance is multi dimensional, in Conclusions Organisation, Geneva, http:// www.who.int/peh/noise/ fact, at low sound pressure levels The ‘A’ frequency weighting is not guidelines2.html (2002.) the character of the noise (e.g. a scale, it cannot be used to temporal structure and frequency ‘establish a human dose response content) can become, by far, the relationship’ and it does not D. W. Robinson, & R. S dominant factor in the annoyance simulate the response of the Dadson , “A re determination of perception. This was clearly shown human ear. The ‘A weighting’ the equal loudness relations for in research carried out by Scannell should always be described in a pure tones”, Brit. J. Appl. Phys. 7, [6] where subjects compared a low glossary as the ‘A’ frequency 166 181 (1956). frequency repetitive impulse noise weighting to distinguish it from a to pink noise for both loudness time weighting. The ‘A’ frequency K. Scannell , “Human response to and annoyance. Here a character weighting must be used for a low frequency, repetitive impulse correction of up to 15 dB was occupational noise assessments but noise”, J. Low Freq. Noise and found to be required where audible should be utilized with extreme Vibr. 8(4), 122 129 (1989). sounds were at a very low sound care when an indication of pressure level but were unpleasant loudness or noise annoyance is Australian/New Zealand in character. required. Standard ‘Occupational noise management’ Part 4: Auditory Scannell found that for annoyance, A possible improved description of assessment, AS/NZS 1269.4 any penalty added to the objective the ‘A’ frequency weighting is: the (1998). o measurement for a source with ‘A’ frequency weighting is used as a unpleasant character must be level rudimentary approximation to the dependant with a higher penalty subjective human perception of for lower sound pressure levels. loudness at low sound pressure The fact that character is more levels. There is a known important than the sound pressure relationship between the statistical
NEW ZEALAND ACOUSTICS Vol. 16, No. 2 21