Hello all
I'm not fluent in english, so I do not know if the terms i will use
are the correct one. Hope you will understand me ! :o)
I have a digital system plug to a microphone. I would like to measure
sound pressure level :
I compute RMS over 4096 samples (@48 kHz -> 85.333 ms)
RMS= sqrt[ (1/4096) sum( x(n)**2 ) ]
Then I compute LEQ as mean(RMS)
LEQ (1s) = mean(RMS) over 12 RMS samples (12*4096/48000 =1024 ms)
LEQ(15mn) = mean(LEQ(1s)) over 879 samples (12*4096*879/48000 = 900 s)
No problem with this values. I compared them with a sound level meters
and have the same results.
Now I would like to compute peak sound pressure level over 15 minutes
I don't now If I should take the max of RMS, the max of my LEQ(1s) or
something else.
The problem is that when I take max of the RMS, I get a very high
value. The sound level meter doesn't give the same value. (The value
is expressed as xxx dBSPL)
I do not filter RMS (do not apply time-weighting). Should I perform
that to compute the peak sound pressure level ? If so what should be
the time constant 125ms or 1s ?
Thank you for your advice,
Curl.
Noral D. Stewart - 20 Dec 2003 01:51 GMT
First, we need to figure out if you are really interested in a peak value or
in a maximum value. These have distinct different meanings when applied to
sound levels.
The peak level is an instantaneous maximum value of the pressure pulse or
wave with no averaging of any kind. Recognizing practical limitations, a
minimum rise time of 50 microseconds is usually required. This kind of
measurement is usually applied to pulses such as a gunshot or hammer blow
and not to continuous sound.
A maximum sound level refers to the maximum level measured with a specific
exponential time weighting such as the standardized fast and slow, or to the
maximum value of a short LEQ sample of specified length. Any time you
specify a maximum sound level, you also have to specify the time weighting
or the length of the LEQ sample. Thus, your maximum one second LEQ value
would be one type of maximum sound level.

Signature
Noral Stewart
www.sacnc.com
> Hello all
>
[quoted text clipped - 26 lines]
> Thank you for your advice,
> Curl.