Pink noise generator

Pink noise generator

Postby Testman » Thu Oct 03, 2013 11:22 am

Pink noise generator


First, thank you for this great software! I really like it, a real bang for your buck. I bought the RTA-168 package primarily for room acoustics measurements and there are two bugs that I want you to know of:
1) The pink noise generator seems to be a bit flaky. Spectrally it looks OK but the RMS value is unduly changing over time so that I can't get a stable dB reading on the multimeter. White noise is perfectly stable (RMS-wise) and that's what I expect from pink noise as well. You can hear by ear that the level is shaky and it's on the test tone CD as well meaning the problem is inherent in the pink noise generator.

2) If I switch the multimeter to dB(A), dB(B) or dB(C) units the numbers refresh with a small glitch (maybe jumping back to 0.000 for a few milliseconds). This glitch is enough to make it very annoying to work with the multimeter as the alignment of the numbers jump back and forth. This same behavior is present in XP and Vista but only for these 3 dB modes.

I hope you can fix these issues soon as they diminish the usability of this otherwise great software for room acoustics quite a bit.

Post at 24-1-2013 21:55

Thank you for your comments.

(1) Pink noise has more energy per Hz at low frequency. If you use a multimeter to measure it, the RMS value may vary quite a lot reading by reading depending on the integration time of the multimeter. If you use the oscilloscope of MI to measure the RMS value of the generated pink noise and set the sweep time of the oscilloscope to 1s, the RMS value measured will be quite stable. If you further increase the sweep time, it will become stabler. If you add a 20Hz high pass digital filter in the oscilloscope of MI (right click anywhere within the oscilloscope and then choose [Oscilloscope Processing]>"High Pass"), the RMS value will become almost stable. Therefore, the fluctuation of the RMS value measured by the multimeter is caused by the low frequency components (<20Hz) of the poink noise which are supposed to have more energy than those high frequency components according to the pink noise theory. The pink noise generated by MI is not band-limited in its digital form, meaning that its energy spreads from 0 Hz to 1/2 of the sampling rate. Of couse, it will be band limited by the sound card when output due to the sound card's limited bandwidth (normally a few Hz to 20kHz) . The residual low frequency components (<20Hz) are outside the audible range and will not affect the equalization accuracy.

On the other hand, it is possible for MI to generate a band-limited pink noise. The steps are as follows:

1. Save a pink noise with a certain length to the hard disk as a wave file using the signal generator of MI
2. Use [File]>[Open] to load the pink noise wave file into the oscilloscpe of MI.
3. Set a proper digital filter in the oscilloscope of MI to band limit the pink noise.
4. Generate the band-limited pink noise using the "Play" or "Cyclic Play" button in the Instrument toolbar of MI directly or save it as a wave file into the hard disk (don't forget to tick the "persist" option in the Oscilloscope Processing page) so that you can play it using other wave file player or even burn the file into an audio CD. You will find that the band limited pink noise will not sound flaky.

(2) Thank you very much for spotting the transient display of "000.0" for dB(A), dB(B), dB(C). These three values come from the spectrum analyzer. "000.0" will be displayed temporarily before the spectrum analyzer finish the calculation of dB(A), dB(C) and dB(C). It is an imperfection/bug and we have fixed it in MI3.3. This issue will not effect the pink noise generation, though.

Post at 19-3-2013 14:35
Posts: 48
Joined: Wed Oct 02, 2013 9:51 pm

Return to Bug Reports

Who is online

Users browsing this forum: No registered users and 4 guests