A-weighting is a frequency dependent curve (or filter) which is applied to sound pressure microphone measurements. A microphone detects sound equally well at all frequencies, but the human ear does not detect sound equally well at all frequencies. The ear detects frequencies between 1000 Hertz to 4000 Hertz well, but other frequencies not as well (below 1000 Hz and above 4000 Hz). Applying A-weighting to a microphone recording attempts to introduce the human perception of sound to the microphone recording.
Article contents: 1. What is A-weighting? 2. The A-weighting Curve 3. How is the A-weighting curve used? 4. Is dB(A) always lower than dB? 5. Table of Values 6. History 7. Other Types of Weighting 8. Simcenter Testlab: Applying A-Weighting
1. What is A-weighting?
A-weighting is a frequency dependent curve (or filter) which is applied to sound pressure microphone measurements to mimic the effects of human hearing.
Given the same sound pressure levels, microphone recordings can be very different than the levels perceived by the human ear (Figure 1).
Figure 1: A microphone (left) and human ear (right) will record/perceive sound differently.There are several reasons why there can be difference in sound levels between a microphone recording and human ear perception of sound.
The sound pressure on the microphone diaphragm versus the ear drum (Figure2) can be very different, even in the same sound field:
The air volume in the ear canal has a resonance around 4000 Hz, causing higher sound levels on the ear drum than the microphone diaphragm
The presence of human head, torso and outer ear interfere with and alter the sound field
Other differences are due to the cochlea hearing organ (Figure 2) and psychological effects:
The hearing organ, or cochlea, has difficulty detecting sounds at low frequencies and very high frequencies
As humans age, damage to the cochlea causes high frequency hearing loss
The cochlea has a logarithmic shape, causing humans to be better able to distinguish changes in pitch at lower frequencies than higher frequencies
Figure 2: Human Ear: Pinna, Ear Canal, Middle Ear, Cochlea
Unlike the human ear, a microphone is not surrounded by a torso, pinna, and ear canal. There is no logarithmically shaped hearing organ in a microphone. It is no wonder there are differences between sound perceived by the human ear and a microphone recording!
2. The A-weighting Curve
A-weighting is an established, standard curve that attempts to alter the sound pressure levels of recorded by a microphone measurement to more closely match the perception of the human ear.
The A-weighting curve (Figure 3) shows decibels of attenuation or gain at every frequency over the range of human hearing. This gain/attenuation is applied to the microphone measurement.
Figure 3: A-weighting curve - dB of accentuation vs frequency
Typically, this gain/attenuation can be applied via:
Analog filter built into a electric circuit in the recording device
Digitally applied after the recording is stored on a computer
Some key features of the A-weighting curve:
Below 1000 Hz, the sound levels are attenuated
At 1000 Hz, there is no gain or attenuation
Between 1000 and about 6000 Hz, the levels are amplified a few decibels
At about 6000 Hz and higher, the sound levels are attenuated
Sound calibrators often generate tones at 1000 Hz to take advantage of the characteristics of the A-weighting filter. The calculated calibration values cannot be affected by whether the A-weighting filter is present or not. Operators cannot accidentally forget to turn on/off the A-weighting filter and unintentionally change the calibration value.
3. How is the A-weighting curve used?
Suppose you measure 80 dB with a microphone at 100 Hz. If you look at the A-weighting curve at 100 Hz, an attenuation value of -20 dB is indicated (Figure 4).
Figure 4: An -20 dB attenuation at 100 Hz
Subtracting 20 dB from 80 dB yields 60 dB. To indicate that A-weighting was applied to the signal, the recorded value would be annotated as 60 dB(A) or 60 dBA.
To remove any doubts, it is also customary to annotate the original, unweighted values as “Linear”. For example, “80 dB Linear” or “80 dBL” or “80 dB(L)”.
This procedure is done at every frequency in the spectrum. It can also be applied to an octave spectrum as well.
4. Is dB(A) always lower than dB?
The A-weighting curve attenuates a great deal of the frequency range. Intuitively, one might guess that the overall dB(A) value will always be lower than the original overall dB (or dBL) value. This is not always the case.
For example, observe the spectrum shown in Figure 5. The overall dB value is 84.2 dB without A-weighting applied.
Figure 5: Spectrum of human whistle, no weighting applied
Now observe the spectrum shown in Figure 6 which has A-weighting applied. The overall dB value is 85.1 dB(A).
Figure 6: Spectrum of human whistle with A-weighting applied
The A-weighted dB value of 85.1 dB(A) is higher than the Linear weighted value of 84.2 dB.
Why? This is because the whistle, with a frequency of approximately 1600 Hz, is in the frequency range (1000 to 6000 Hz) of the A-weighting curve where sound levels are amplified. The net effect of this amplification is greater than the reduction in the spectrum at all other frequencies. For example, the levels below 1000 Hz are greatly reduced by introducing A-weighting, but because they are lower in amplitude their net effect on the overall level is less than the amplification of the 1600 Hz whistle.
5. Table of Values
Table 1 contains a list of the A-weighting attenuation and amplification values in dB for 1/3 octave center frequencies.
Table 1: A-weighted attenuation values in dB vs Frequency
The A-weighting curve was derived from the Fletcher-Munson curves of equal loudness. The American National Standards Institute (ANSI) was the first to implement the curve in Sound Level Meter standard published in 1936.
It should be kept in mind that the curve is not a perfect representation of effects of human hearing. In order to be able to develop cost effective analog circuits, the curve had to be simplified. For example, the A-weighting curve does not change as a function of the sound level like human hearing.
7. Other Types of Weighting
In addition to A-weighting, there are other acoustic weighting functions. They include B, C and D weighting as shown in Figure 7.
Figure 7: A, B, C, and D Weighting Curves
Used in various applications and industries, the B and C weighting curves are similar to A-weighting, but do not have as much attenuation below 1000 Hz. The C weighting curve is the flattest of the A, B and C curves. The D-weighting curve is typically used in very high pressure aeronautical noise applications, like airplane flyover noise.