Skip to main content

Select your region & language

Global

Region

octave

Oto-kun
Dad, when you talked about Pascal and decibels the other day, you told me that the unit for noise level is dBA, right? Today, while I was waiting for the bus at the station, there was a display on the tower in the green space in front of the station that showed the current noise level in dB. Since it's a noise level, I thought it might be A-weighted like you taught me, but the display didn't have an 'A' on it.
father
The term 'noise level' refers to the A characteristic, so it was displayed without the A. General people might think that decibels are somewhat a unit of sound, but if it has an A, they might wonder what it is. In the past, a unit called 'phon' was also used, but since 1997, it has been standardized to dBA. The term 'noise level' is used only in Japan, and the precise term is 'A-weighted sound pressure level.'
Oto-kun
It's hard to remember a long name. 'Noise level' is clearer and better.
father
Noise levels are actually very convenient. Noise, for the most part, has a complex range of frequency components from high to low sounds, and it also fluctuates significantly in intensity over time. The fact that we can evaluate human responses to this complex noise with a single number is quite remarkable.
Oto-kun
A while ago, you told me that human sound perception includes frequency masking and temporal masking. Do noise levels also include those effects?
father
That's not correct. There is a measure called 'loudness' that represents the intensity of sound, but this loudness takes masking into account and accurately captures the perceived loudness of sounds by humans. However, the calculations are complex, and there are still not many measuring instruments available in the world to measure this. Recently, loudness has started to be used in evaluating noise from office equipment and the sounds inside vehicles, but road traffic noise is often assessed based on noise levels.
Oto-kun
Why is that?
father
When machines are not functioning properly, they can sometimes be identified by the sounds they make. The sounds that occur in abnormal states are referred to as 'abnormal sounds,' but when evaluated at noise levels, they often do not accurately represent the loudness perceived by humans. This is because abnormal sounds often consist of dominant frequencies, and by considering frequency masking, the perceived loudness of the sound can be represented. Conversely, road traffic noise includes a wide range of sound frequencies, allowing it to be evaluated by noise levels.
Oto-kun
Is it the frequency range? Depending on the type of noise, it is necessary to decide how to evaluate it.
father
That's right, I remember when we talked about frequency masking, I explained the vibrations of the basilar membrane within the cochlea. Human hearing generally has an A characteristic, but in reality, we have filters to distinguish between different pitches. The bandwidth of these filters is called the critical band, which is about 1/3 of an octave.
Oto-kun
??? I don't really understand. First, human hearing has filters that distinguish between high and low sounds. ..1/3 octave? ..what does that mean?
father
You know about octaves, right? An octave spans from one low C note to the next C note. In terms of frequency, this octave is double. The sound of the time signal is based on the note A, but the first three notes are at 440 Hz, and the last long note is the A an octave higher at 880 Hz.
Oto-kun
Which note on the guitar corresponds to 440 Hz?
father
It's the sound of the 5th fret on the 1st string. On a guitar, the 1st fret is a half step, so the next A note is at the 17th fret of the 1st string. In other words, a half step is 1/12 of an octave, since there are 12 half steps in an octave. On a piano, adjacent notes, including black keys, are also half steps, so they are similarly 1/12 of an octave.
Oto-kun
So, a 1/3 octave consists of 4 half steps, which equals 2 whole steps, making the interval between C and E, right?
father
That's right. An octave is a doubling of frequency, but in sound analysis, a central frequency based on a neat 1000 Hz is used instead of do-re-mi. Frequencies like 63 Hz, 125 Hz, 250 Hz, 500 Hz, 1000 Hz, 2000 Hz, 4000 Hz, and 8000 Hz are used, for example. I previously mentioned that humans can generally hear sounds ranging from about 20 Hz to 20 kHz. However, the sounds present in our living environment typically range from about 63 to 8000 Hz, so it is common to evaluate sound using around these eight bands.
Oto-kun
What is a band?
father
Oh, I haven't talked about the band. A frequency band with a width of one octave is called an octave band. The calculation is complicated, so
I won't explain it in detail, but an octave band with a center frequency of 1000 Hz has a width of 770 to 1410 Hz, and an octave band with a center frequency of 500 Hz has a width of 355 to 710 Hz. The sound intensity level contained in a one-octave band is called the octave band level. When analyzing sound, it is easiest and most straightforward to evaluate the noise level with just one number in dBA. However, with the noise level, you can't tell whether the noise has a lot of low-frequency sounds or high-frequency sounds. Therefore, by analyzing with octave bands or even more finely with 1/3 octave bands, you can capture the frequency characteristics of that noise.
Oto-kun
But what is the purpose of conducting such an analysis? If there are no abnormal sounds, the evaluation can be based on the noise level, right?
father
Consider taking measures against noise. By understanding the levels by frequency, you can determine which frequency of noise needs to be addressed. For example, when measuring road noise, having not only the noise level but also the octave band measurements allows you to design sound barriers optimally and estimate the noise levels in living spaces, such as homes close to the road.
Oto-kun
I see. While the noise level is sufficient for evaluation, considering countermeasures and predictions, octave band analysis is effective.