> The A-law issue is actually more complex than just 'levels'.
I agree with you, it certainly is. And I particularly agree with the idea =
that
the use of a weighting curve to create a single number noise level doesn't =
tell
the whole story.
But the most recent work on equal loudness, as embodied in the ISO 226: 200=
3,
shows that A weighting does in fact conform quite well to equal loudness
contours, at least relative to ITU-R BS.468. The data in the fourth figure=
from
your link to the Wikipedia article on A-weighting illustrates that.
Going back to the BBC research report EL-17, their emphasis was primarily o=
n
types of noise other than the random noise produced by microphones. They w=
ere
attempting to find a noise measurement method that produced ratings equal t=
o
subjective ratings for impulsive noises, carrier leakage, hum, and other ty=
pes
of noise. So EL-17 isn't really relevant to the question at hand regarding=
self
noise of microphones.
So what is relevant to microphone noise measurement? Noise measurements fo=
r
microphones had ought to allow us to make useful comparisons between one ty=
pe of
microphone and another. For instance, if one microphone has a measured sel=
f
noise that is lower than another, then it had ought to sound quieter. Does=
A-weighting do that? Does CCIR weighting? If we were to order the noise
measurements of microphones, first using A-weighting and then using CCIR
weighting, then which one creates an order that is most like subjective ran=
king?
I don't have an answer to that, but it turns out that rankings using
A-weighting and CCIR weighting are very close. If one subtracts the A-weig=
hted
self noise from the CCIR-weighted self noise, the difference is almost alwa=
ys
between 9 and 11 dB. So both weighting systems are almost equivalent for t=
hat
purpose. My audio analyzer allows me to easily select either weighting fil=
ter,
but I stick with A-weighting because there is very little effective differe=
nce
between the two systems for my purposes.
One method that would be better would be to use a loudness measuring system=
that
mimics the way human hearing assesses loudness. One such method is that
proposed by Moore and Glasberg:
Moore, B.C.J., Peters, R.W. & Glasberg, B.R. (1996) "A revision of Zwicker'=
s
loudness model", Acta Acustica, vol.82, pp.335-345.
These types of models are complicated, but they produce results which are m=
uch
more nearly perceptually correct than do simple weighting networks. I'll w=
ork
up some numbers if anyone is interested.
Meanwhile, does anyone know of manufacturers who publish both A-weighted an=
d 468
weighted noise specs, other than AKG, Sennheiser, Schoeps, and DPA?
|