The "A" in dB(A): A-weighting
A sound level meter is not the human ear. Two basic components of a modern sound level meter (like the one above) are a microphone and a computer, which is programmed to interpret - as sound levels - the minuscule rapid changes in air pressure detected by the microphone. This is the engineering approximation to the human ear (microphone) and brain (computer). The A frequency weighting (A-weighting) was developed as a way to account for human sensitivity to different sound frequencies so that a sound level meter can better "hear" like a human. Also, A-weighting can help to compare some sounds that have different frequency content using a single-value parameter.
Three things to know:
- 1. Why was A-weighting created?
- 2. How does one measure a dB(A) sound level?
- 3. Is a dB(A) is a dBA is a dBa always a dB(A)?
The Reason for A-weighting
We humans have various sensitivities to different sound frequencies. What this means is that some sound frequencies must be louder than others for them to sound like they have the same loudness (volume). We hear best at frequencies between approximately 500 and 8,000 Hz. This range of frequencies lies within the frequency range of human speech. For frequencies below 500 Hz (oftentimes what is meant by “low” frequencies), our sensitivity diminishes rather quickly. For example, a 20 Hz sound would need to be 47 decibels greater (louder) than a sound at 500 Hz in order for the two to sound like they have the same volume. We do not hear very low frequency sound very well, that is, unless the sound (decibel) level is very high.
How a Sound Becomes a dB(A)
The following describes the basic process of how a sound in air becomes an A-weighted sound level on a sound level meter: A microphone diaphragm moves back and forth in response to the sound pressure. The movement of the microphone is detected in the sound level meter as an electrical voltage. The voltage is then converted to pressure according to the microphone's calibration sensitivity (e.g. 50 mV/Pa). The pressure is then converted to decibels relative to a reference value (e.g. 20 μPa). The signal is processed to determine the frequency content, that is, the sound level per octave or sub-octave band frequency. At this point, the un-weighted sound pressure levels per frequency band are known. Then, A-weighting is applied to each frequency band according to prescribed standards (e.g. ISO 226:2003); the specified decibel weighting factors are added to the un-weighted decibel values (note that most of the A-weightings have negative values). Finally, the frequency bands are logarithmically summed to yield the total, A-weighted sound pressure level, expressed as dB(A).
Notes on Nomenclature and Usage (User Beware)
The 'A' in parentheses after dB indicates that A-weighting has been applied. The unit "dB" without the (A) indicates that no weighting has been applied, so the level is un-weighted (also called "linear"). Sometimes, dB(A) appears as dBA or dBa, or some other variation, which is fine (although lowercase ‘a’ or ‘b’ is not preferred). However, as a word of warning, sometimes "dB" is erroneously noted or listed when the level represented is actually A-weighted. This can be a problem. A-weighting is so prevalent for sound measurements across acoustical sectors (environmental noise, industrial hygiene, product noise, etc.), that it is sometimes omitted, being taken for granted that the “(A)” is implied. This can be disastrous in some cases if not caught, especially if the values differ significantly between the linear (dB) and A-weighted (dB(A)), which is often the case. One must carefully observe units, labels, and footnotes, and always question whether the correct weighting has been noted.
Another point that should be made is a single dB(A) value (in the absence of any specified frequency information) is typically understood to represent the total A-weighted sound level, that is, the sum of all A-weighted frequency bands. This does not necessarily have to be the case; individual frequency bands can be listed as A-weighted. When spectral values are provided and used for design purposes, it is critical to ensure that the correct weighting is specified. Industry standard practice is for linear octave bands to be presented with the total A-weighted value. Unfortunately, exceptions abound.
There is much more to learn about A-weighting such as its limitations, the contexts in which it is used, and how one dB(A) value might compare to another and in what manner. Also, it is useful to know how a dB(A) value compares or relates to other acoustical measurement parameters and acoustical descriptors such as Insertion Loss (IL), Transmission Loss (TL), Noise Criteria (NC), Room Criterion (RC), Room Noise Criteria (RNC), and a bevy of others.
A-weighting is a method to adjust the frequency content of sound levels to approximate the way humans perceive sound. A-weighting accounts for the fact that we hear different frequencies with relative loudness. We are least sensitive to low frequency sounds. One must be careful to note which weighting was used when reviewing submitted information and when providing acoustical data.
About the Author - Tim Simmons, Ph.D. - USSI Director of Industrial Acoustics
Tim Simmons manages USSI's Acoustics Division. Tim comes to USSI with an immense background and knowledge in Acoustics and Noise Control, including software noise modeling. Tim holds a PhD in Physics and Astronomy from the University of Mississippi; and a B.S. in Engineering Physics from the University of Tennessee.