Gentle people:
The A-law issue is actually more complex than just 'levels'.
One=C2=A0issue here is that there are quite possibly an infinite number of =
shaped
noise spectra that can register as the same A-law level. So if the self noi=
se of
the mic is measured as being 16 dBA, the noise spectrum of that mic=C2=A0co=
uld favor
the low, the mids or the highs and still be read by the sound level meter a=
s 16
dBA.
The Fletcher-Munson study that gave rise to the 40 phon (in level) A-law cu=
rve
was done a long time ago and was=C2=A0formulated to account for the percept=
ion of
individual tones, not random noise, back in 1933. Go to the audiologist
and=C2=A0that's what you get: tones to test your hearing, not=C2=A0even ban=
d-passed
noise!=C2=A0
In particular, A-law weighting of low-level background noise (such as the d=
rone
of the city) does NOT correlate at all well with what most people hear, it=
devalues=C2=A0the low end perception of the spectrum=C2=A0severely. And A-l=
aw weighting
does not account for our enhanced perception of low-level noise at=C2=A0aro=
und 6=C2=A0KHz;
it devalues it by 10 - 12 dB. That is why the BBC and others did the resear=
ch to
account for this perception and=C2=A0formulated the ITU-R 468 spec. Note th=
at this
spec specifies not only the weighting curve, but also the precise method fo=
r
metering (measuring) the level=C2=A0of the noise.
ITU-R 468 is better (and much newer!) science, not an attempt to hammer
something into a 'compartment' in which it does not belong.
Here are some links:
http://en.wikipedia.org/wiki/Equal-loudness_contour
http://en.wikipedia.org/wiki/A-weighting
http://www.rane.com/par-w.html#weighting_filters=C2=A0 (look towards the en=
d for
different weighting filters)
http://sound.westhost.com/project17.htm
http://broadcastengineering.com/mag/broadcasting_reading_microphone_specifi=
cations/
I can see why Jez has an issue with A-law specs for mic self-noise. But is =
ITU-R
468 the be-all and end-all spec? For some people, maybe not, for their own=
perception may be heightened or stunted in ways different from the 'norm' o=
f the
human population. But I do believe that applying A-law curve, which was des=
igned
for the perception of tones, is not applicable to measuring noise in
any=C2=A0reasonable way. Why folks continue to apply it to such purposes is=
quite
puzzling. Perhaps they didn't know before ( in the '60s) what we know now a=
nd
there is just too much inertia to move away from A-law. It's sad that the m=
ic
manufacturers haven't kept up with current perception theory and measuremen=
t
methodology.
Human perception, especially hearing, is quite complex. Pick up a contempor=
ary
psychoacoustics text and you'll see.
ITU-R 468 is much better than A-law for self-noise measurement, but it's st=
ill
an approximation to what really happens in our minds when we listen to nois=
e and
make our own judgements about it.
So what do we do now?=C2=A0 :>}
Numbers can lie. I suggest we listen and make more comparison recordings of=
different mics on a common (good) recorder under identical extremely quiet=
conditions. And by identical, I mean not only the same background level, bu=
t
with the preamp gain calibrated to give the same output level for each mic=
according to a properly chosen calibration tone. Then we can make our own
judgements independently from the vendor's 'spec' sheets.
Let's move forward.
- Greg
|