| Japanese | English |

AUTOMATIC MEASUREMENT AND IDENTIFICATION BY EA
(Analysis of Toilet Flushing Noise 2)

It was measured using two personal computers. SONY VAIO PCG-505 R/DK was used for reproduction and Dell Inspiron7500 was used for measurement. Measurement software is DSSF3.

The water of a toilet is poured 2 seconds after the measurement starts. The valve closes at 10 seconds. There is a quite loud sound. After that, the sound of which water is supplied continues till the measurement ends.

Time change of Φ(0) (sound pressure level) in SA. Click the Play button in this window to reproduce a recorded sound. Sound is recorded using the microphone MS907 of SONY, and it is measured by EA (Environmental noise Analyzer) automatically.

This is the main window of EA.

It is the setting screen of the input device. This screen is displayed when you click the mark of the microphone in the tool bar, or choose the "Input device settings" from the pull down menu. Here, the microphone MS907 is chosen.

Click the Edit button to confirm the calibration. The input level should be adjusted for the sound level meter reading.

This is the setting window of the measurement conditions. Sampling rate should be set as 48 kHz. Trigger SPL (20 dB in this example) is set as the relative level.  It means that the measurement starts when the SPL exceeds 20 dB from the beginning (SPL for first 2 seconds is set as 0 dB). Time constant is a duration in which the SPL is averaged. Measurement timing is a setting to specify the start timing of the measurement. It can be chosen between the Maximum level and the Trigger SPL.

Detailed explanation of a setup of the Running ACF is omitted here. See the program manual. Integration interval is set as 0.02 sec. This is for capturing the change of the sound pressure level precisely. Identification timing was set as Maximum SPL tentatively.

This is the peak level monitor. The signal level is displayed relative to the peak level of the built in sound circuit.

This is the SPL meter. Sound pressure level is displayed in dB. Trigger SPL is monitored by this value. The microphone must be calibrated by using the sound level meter or the acoustic calibrator.

How to make a noise source template.

Here, the previous (Analysis of Toilet Flushing Noise 1) analysis result is explained again.

The t₁ is the time delay of the first peak of the ACF that is the reciprocal of the fundamental frequency. This value is 0.8 ms when the toilet is flushed and is 0.4 ms when the water is supplied. The f₁ is the amplitude of the ACF peak, that corresponds to the perceived pitch strength. It was found that this value is large when the sound level is high. Background noise level was 38 dB and the maximum noise level was 68-70 dB.

Noise source template is created.

Sound level (Phi(0)) is set as 68 dB and t₁ is set as 0.8 ms. Weightings for them are set as 1. Also, t_e is set as 4 ms and Phi1 is set as 0.1. Weightings for them are set as 0.5. Name this file as "wash" and click the New button to save a new template. Since the sound pressure level of Toilet Flushing Noise is almost the same each time, and there is also information on t₁, f₁, and Taue, Toilet Flushing Noise should be identified correctly.

The sound was reproduced by VAIO, and the automatic measurement was tested by Dell Inspiron7500 with the above setup. It has identified satisfactory. Measured sound pressure level (61.66dB) is close as a template (68dB). t₁ was 0.79, and f₁ was  0.82. t_e was 17.07ms.

Test was repeated ten times. Sound level was between 60 and 66 dB. To raise identification accuracy more, it is necessary to set the time of t_e minimum as identification conditions. t_e becomes the minimum at the moment a signal changes. So t_e minimum is used  to compare the state of the signal. Then, t₁ should be 0.4 and f₁ should be 0.16-0.3.

Measurement condition was changed. Running step was set as 0.01 sec. Identification is set as Minimum t_e.

Same template is used.

As the result of the automatic measurement, SPL were around 67 and 68 dB, Tau-e were 6.56, 6.82, 6.54, 7.45, and 10.4, t₁ were 0.35 for all times, f₁ were 0.42, 0.14, 0.13, 0.15, 0.22, and 0.26. It was almost  a same result as I thought.

Template is changed.

Sound file of the flushing noise
wc2.wav (1,551KB)

Click the above file to play a sound. You can also download the file. Then, it can be loaded by RA and analyzed by SA as in this page.

December 2002 by Masatsugu Sakurai