Technical Report: What is a Sound Level Meter? 11

Table 11-1 Classification of frequency analysis

	Analysis width	Main uses
FFT method	Constant width analysis	Investigation of physical causes and countermeasures
Octave Analysis	Constant ratio width analysis	Noise and vibration evaluation

Octave analysis is also known as CPB (Constant Percentage Bandwidth) analysis. This section explains octave analysis.

To implement noise reduction measures, noise analysis, or frequency analysis, is necessary. Octave analysis is a long-established method of noise analysis. An octave refers to twice the frequency (see "What is an octave?" below).

Octave analysis is frequently used because the frequency response perceived by the ear is geometrically uniform. For noise being measured, the sound pressure level for each band is determined by passing it through a bandpass filter specified in 1/1 octave or 1/3 octave standards within the audible frequency range. The characteristics of the filters are specified in JIS C 1514:2002, so please refer to that for details.

What is an octave?

An octave refers to a musical interval where the frequency ratio is double, such as the relationship between a C note and the next C note above it. An octave band is a frequency range (band) where the ratio of the upper and lower frequency limits is one octave, and the frequency at the center of this band is called the octave band center frequency. An octave band divided into thirds is called a 1/3 octave band. As shown in the diagram on the right, the equal temperament interval in music is an interval obtained by dividing one octave into 12 equal parts proportionally, resulting in a 1/12 octave band.

The JIS (IDT) translation of the IEC standard specifies filter characteristics, including not only octave and 1/3 octave filters, but also more generally 1/N octave filters. It also specifies tolerances for required accuracy up to two classes (1 and 2). It includes requirements not only for filter shape but also for equipment (such as the effects of environmental changes and EMC requirements). This standard requires octave frequency ratios that are powers of 10. The structure of the standard is similar to that of the JIS C 1509 series sound level meters.

Table 11-2 below lists the center frequencies of octave and 1/3 octave filters. Note that these center frequencies are nominal center frequencies as defined in ISO 266; the exact center frequencies are calculated using the formula specified in standard 1.

center frequency (Hz)	1/1 octave	1/3 octave	center frequency (Hz)	1/1 octave	1/3 octave	center frequency (Hz)	1/1 octave	1/3 octave
0.8		○	25		○	800		○
1	○	○	31.5	○	○	1000	○	○
1.25		○	40		○	1250		○
1.6		○	50		○	1600		○
2	○	○	63	○	○	2000	○	○
2.5		○	80		○	2500		○
3.15		○	100		○	3150		○
4	○	○	125	○	○	4000	○	○
5		○	160		○	5000		○
6.3		○	200		○	6300		○
8	○	○	250	○	○	8000	○	○
10		○	315		○	10000		○
12.5		○	400		○	12500		○
16	○	○	500	○	○	16000	○	○
20		○	630		○	20000		○

Equation 11-1

Equation 11-2

Table 11-3 Strict center frequencies of 1/3 octave bands in the audible range

index x	Nominal center frequency (1/3 octave) [Hz]	Strict center frequency based on powers of 10 10 ^{x / 10} x 1000 [Hz]	Strict center frequencies based on powers of 2 [Reference] (Filter design based on powers of 2 is not recommended in JIS C 1513-1.) 2 ^x/3 x 1000 [Hz]
-16	25	25.119	24.803
-15	31.5	31.623	31.250
-14	40	39.811	39.373
-13	50	50.119	49.606
-12	63	63.096	62.500
-11	80	79.433	78.745
-10	100	100.00	99.213
-9	125	125.89	125.00
-8	160	158.49	157.49
-7	200	199.53	198.43
-6	250	251.19	250.00
-5	315	316.23	314.98
-4	400	398.11	396.85
-3	500	501.19	500.00
-2	630	630.96	629.96
-1	800	794.33	793.70
0	1000	1,000.0	1,000
1	1,250	1,258.9	1,260
2	1,600	1,584.9	1,587
3	2,000	1,995.3	2,000
4	2,500	2,511.9	2,520
5	3,150	3,162.3	3,175
6	4,000	3,981.1	4,000
7	5,000	5,011.9	5,040
8	6,300	6,309.6	6,350
9	8,000	7,943.3	8,000
10	10,000	10,000	10,079
11	12,500	12,589	12,699
12	16,000	15,849	16,000
13	20,000	19,953	20,159

To convert the dB values of known 1/3 octave data to the corresponding 1/1 octave band data, calculate the sum of the dB values of the 1/3 octave band data corresponding to the 1/1 octave band you want to obtain. For example, when obtaining the band data value for a 1/1 octave with a center frequency of 1000 Hz, if the corresponding 1/3 octave band data has the following dB values: