Hi--
Thanks for the interesting thoughts about a method to employ just
ones recorder and mics. I guess the method depends on whether Pratap
needs the usual, A-weighted measurements or consistent numbers for
personal use.
With the same record level and mics one generates consistency that
can be used in many ways. I can measure moments of presence with RMS
averaging and I even can "see" when significant differences occur in
the wave form display because my file saturations are high enough.
What I'm personally curious about is being able to translate RMS
A-weighted measurements in my DAW to dBA units because these numbers
match most of the other available references.
It makes sense to use both a 1K tone and band limited pink noise when
running the experiment. Is there an opinion about whether the
speaker should be set-up at 1 meter? Below are updated steps. Is the
math as simple as I describe in Step G? Further corrections and
additions _gladly accepted_.
(A) Outside, at least 6' above open ground, play band-limited pink
noise, 200 Hz to 2 kHz from a speaker so each reads 64 dBA on a SLM
at 1 meter. (1K Hz step below)
(B) Set-up your field mics as physically close to the sound level
meter as possible and direct one of the mics at the speaker. Note
which channel was directed at the speaker.
(C) Set the preamp gain of your recorder (low cut off) at your usual
setting (must be a setting you can repeat).
(D) Make a recording of the pink noise reference sound.
(E) Play the 1k Hz reference tone and adjust speaker playback level
so the SLM reads 64 dB. Make a recording of this reference sound.
(F) Play each of the test types on your editing app and use A-weighted
measurement to determine the saturations the 64 dBA stimuli produced
in the recording .
(G) With these numbers you can use simple math to extrapolate
approximate dBA measurements calculated as -20 dBA plus the -
(negative) dBA amount subtracted from the 64 dBA reference level.
For example, if the pink noise measures -20 dB with A weighting in
your DAW, this means a moment of background presence that measures
-20 dBA is equal to 64 dBA. Thus, a location recording with a 30 dBA
ambient sound level should produce an A weighted DAW measurement of
-54 dBA, or -20 + -34 dBA (64-30).
I guess we'll see if the 1000 Hz tone produce a different RMS A
measurement and explore the significances. Rob D.
= = =
At 1:03 PM -0800 11/27/10, Gregory O'Drobinak wrote:
>Hi!
>
>I find it rather peculiar that folks are worried about a 2 dB gain
>error due to
>complications using a single 1 KHz tone for doing the calibration as Raimund
>suggested and yet are not worried about the huge calibration error
>due to using
>the band-limited pink noise (BLPN) as a stimulus!
>
>By using a single 1 KHz tone for doing the calibration, both your recorder and
>the SLM (sound level meter) are reading the same energy and so the calibration
>is valid.
>
>But by using the BLPN, the SLM is integrating that sound energy *after* the
>A-weighting filter has been applied. That means that energy around 200 HZ is
>attenuated by ~10 dB, so all of the energy gets progressively
>"discounted" below
>1KHz by the weighting filter.
>
>When you use that same BLPN stimulus with your recorder, it does not have that
>same A-weighting filter built in, so its level meter integrates the
>sound level
>energy *flat* down to the 200 Hz edge of the stimulus. So your
>"calibration" of
>that recorder is off relative to the energy at 1 KHz, given A-weighting, and
>this calibration point is only valid for only *one* stimulus spectrum, namely
>that of a 200 Hz - 1KHz BLPN. I can come up with an infinite number of
>band-limited noise stimuli that have different spectra, but will read exactly
>the same as a 200 Hz - 1KHz BLWN stimulus, on the SLM using dBA.
>This is a very
>important concept and is one reason why other systems for measuring loudness
>have been developed. I don't believe that A-weighting is the holy grail of
>loudness measurement.
>
>Anyway, the point here is that your recorder meters level as a linear function
>across the whole frequency band (assuming it really has flat
>response), but the
>SLM meters level according to a non-linear frequency curve that attenuates
>energy quickly below the 1 KHz 0 dB point on the A-weighting curve. Therefore
>using any stimulus other than a pure tone at 1 KHz, which is where both
>units/systems are at the same 0 dB point, will lead to inaccurate
>calibration of
>your recorder.
>
>I suggest comparing the two methodologies. Put the SLM on a tripod and do the
>measurement using a 1KHz one and get your body well away. Use
>A-weighting on the
>SLM. Do the same thing for your mics/recorder using its built-in
>meter. Note the
>calibration point for your recorder.
>
>Repeat the process with the SLM and you recorder using the 200 Hz - 1KHz BLPN.
>Note the results and compare them to the single-tone measurements.
>Then do the measurement using the same BLPN stimulus with the SLM switched to
>C-weighting. Note that difference. Now bring the stimulus level down until the
>C-weighted SLM measurement is the same as the A-weighted SLM measurement. Note
>how much you have attenuated the stimulus to get this last reading.
>It should be
>indicative of the gain difference of your recorder settings produced
>by the two
>methodologies.
>
>Please re-read Raimund's response form today @ 4:44 AM. There is
>much wisdom in
>his response. I especially like his comment about how animals "don't
>care about
>A-weighting at all". The point here, IMHO, is that other creatures
>hear in ways
>much different from our own ears. So if we apply a single artificial loudness
>measurement level to a natural soundspace, what relevance does it really have?
>I believe that the spectral distribution of energy is much more important and
>that there are many spectral distributions that will give the same A-weighted
>level measurement number. For a given number, which is more loud to your own
>ears? You may be surprised.
>
>Check out this paper when you have
>time:
><http://www.envisys.com/us/en/Impact_Sound_Pressure.pdf>http://www.envisys.com/us/en/Impact_Sound_Pressure.pdf
>
>There is nothing wrong with calibrating your recording gear, providing it is
>done in a careful manner. That way you have a benchmark for the level in any
>given setting. Perhaps Bernie can outline his procedure for
>calibrating his mics
>& recording equipment.
>
>Take care,
>
>Greg
>
>________________________________
>From: Rob Danielson <<type%40uwm.edu>>
>To:
><naturerecordists%40yahoogroups.com>
>Sent: Sat, November 27, 2010 11:34:31 AM
>Subject: Re: [Nature Recordists] Re: Responses to ambient sound recording
>
>At 5:06 AM -0500 11/27/10, Dan Dugan wrote:
>>
>>Raimund wrote,
>>
>>> I would suggest to use the sound level meter only for calibrating
>>>your microphone and the recorder. You could for instance record a 1
>>>kHz test signal through a cheap speaker and record it at a
>>>well-defined distance and then measure the sound level by using the
>>>sound level meter at the same distance. You would then know the
>>>absolute sound level of the recorded 1 kHz test signal as well as
>>>the level of the noise that you record at the same gain setting. In
>>>this case it would be of course sufficient to have a simple sound
>>>level meter with a limited frequency range.
>>
>>Correct in concept but one has to be very careful in using a tone.
>>I've tried it with 1 kHz, and found that if I hand-held the meter
>>the reflection from my body will change the reading up or down as
>>much as 2 dB. I suggest using band-limited pink noise, 200 Hz to 2
>>kHz, for the stimulus, at a level of 64 dBA (slow) measured by the
>>sound level meter at the mic location, the SLM on your field mic
>>stand at the height you use. Then put the capsule in the same spot,
>>with field windscreen, and the recorder set to a precisely
>>repeatable record gain.
>>
>>> The final noise measurements could then be carried out directly on
>>>the recorded .wav files.
>>
>>Using an A-weighted software meter for the level calibration.
>>
>>> This would also allow you to determine the spectral distribution
> >>of the ambient noise, which may also significantly influence the
>>>vocalization parameters of the singing birds. A number of studies
>>>have shown that many animal species shift their songs towards
>>>higher frequencies in order to avoid masking effects by
>>>low-frequency ambient noise.
>>
>>Third-octave analysis would surely be enough for this purpose, maybe
>>even octave analysis to reduce the quantity of data to be compared.
>>
>>-Dan
>>
>
>Please correct/add to as necessary:
>
>(A) Outside, at 6' above open ground, play band-limited pink noise,
>200 Hz to 2 kHz form a speaker so it reads 64 dBA on a SLM at _______
>ft distance
>
>(B) Set your mics/recorder at a usual, repeatable record level
>setting in the same spot where the meter was and record the same pink
>noise.
>
>(C) Play the recording on your editing app and use A weighted
>measurement to determine the dBA reading the 64 dBA stimulus produced.
>
>(D) With this number you can perform simple math. For example, if the
>pink noise measures -20 dB with A weighting in your DAW, this means
>-20 dBA is equal to 64 dBA. So a quiet location recording with a 30
>dBA ambient sound level should read as -54 dBA [-20 dBA minus 34 dBA
>(64-30) ] when read using A weighting in your DAW.
>
>Rob D.
>
>= = = =
>
>If any one does this, please share the specifics with us. Rob D.
>
>--
>
>
>
--
|