In the previous article, we introduced an example of pass/fail judgment the quality of motor noise in automotive electrical components. We found that roughness, a sound evaluation index, is useful for evaluating and pass/fail judgment the quality of sound, particularly the "roughness" and "muddiness" characteristics of sounds generated from machinery with rotating mechanisms.
This time, using the same sample, I would like to explain an example of applying fluctuation sound analysis as a useful indicator in the process of developing countermeasures.
The 1/3 octave band analysis and comparison of OK/NG samples for roughness, as shown last time, are reproduced below. The sounds of the OK and NG samples, the same as last time, can also be played using the buttons below.
-
Figure 1: 1/3 Octave Analysis (Blue: NG, Red: OK) -
Figure 2. Roughness spectrum (blue: NG, red: OK)
As shown in this figure, even when differences cannot be clearly extracted using sound pressure level or loudness, it is possible to extract the "roughness" characteristic of the timbre of the motor sound of defective products by using roughness, one of the qualitative evaluation quantities, as an evaluation index. In such cases, it is thought that it may be possible to automate the OK/NG judgment by setting a standard value. Furthermore, it is also possible to make judgments using a combination of indicators, such as sound pressure level, in addition to roughness.
In the column before last, I gave a very brief explanation of roughness and variability intensity as follows:
| Roughness: | This index quantifies the human perception of what constitutes "roughness" in sound (the "roughness" is maximized for sounds that fluctuate at a rate of approximately 70 times per second). |
| Variability intensity: | This index quantifies the human perception of sound as "fluctuating" (the "fluctuating sensation" is maximized for sounds that fluctuate approximately 4 times per second). |
To give you a real auditory experience, we have created the following samples. Please check out the "fluctuation" and "roughness" for yourself.
-
Figure 3 Roughness and Variability
(Clicking on each speaker icon in the diagram will play the corresponding frequency sound listed above.)
As shown earlier, the roughness value is centered around a fluctuation frequency of 70 Hz. For sounds that fluctuate faster or slower than this frequency, the roughness value will be smaller. However, as can be audibly confirmed in the example in Figure 3, the "roughness" is perceived over a fairly wide range of fluctuation frequencies.
The color map of fluctuating sound analysis shown below uses color to represent the magnitude of sound fluctuations, with frequency on the horizontal axis and fluctuation frequency on the vertical axis. Because the magnitude of fluctuations can be grasped not only on the frequency axis but also on the fluctuation frequency axis, it contains more information than roughness and can be used for more detailed sound evaluation and extraction of information that can lead to countermeasures.
-
Figure 4. Results of fluctuation sound analysis (Left: NG product, Right: OK product)
In the example in Figure 4, the frequency band shown in blue on the vertical axis for the defective product corresponds to the band shown as high in the roughness bar graph in Figure 2. While the value at 4kHz is the largest in roughness, the largest fluctuations in the fluctuation sound analysis are at 1.6kHz and the box (orange) with a fluctuation frequency of 30Hz. At 4kHz, the fluctuation sound is strong across a wide range of fluctuation frequencies, whereas at 1.6kHz, the fluctuation is large in a narrower range of fluctuation frequencies centered around the fluctuation frequency of 30Hz.
Noise generated by rotating machinery generally includes fluctuating sounds that depend on the rotational speed. For analyzing abnormal noises like the one in this case, we believe that fluctuating sound analysis, as shown in Figure 4, can provide useful information for countermeasures.
●All data in this measurement column was obtained using our OS-2740 sound quality evaluation pack and OS-2750 variable sound analysis pack.
Ono Sokki- OS-2740 Sound Quality Evaluation Pack (Discontinued)
Ono Sokki- OS-2750 Fluctuation Sound Analysis Pack (Discontinued)
(Excerpt from the email newsletter issued on December 17, 2009)
