OT: S/N-Ratio measurment - help

[Martin Czech]

:::The A-curve may be rubbish, but it is established rubbish, and it is important
:::for us to compare like measurements. I am reminded of the shady practice of
:::spec'ing peak power on amps to make them sound much more powerful then they
:::were. If you compare rms power, the amps sound similar. 
:::
:::The National Semiconductor LM3915 Spec sheet shows a filter for a VU response. I
:::think that is an A-weighted response, but I am not sure.

Of course, for a psycho acoustical meaningful measurement some kind
of "inverse" ear filter will be needed, for example in the case of
determination how loud some industrial noise will be perceived.

I just wanted to mention that the experts seem to disagree what the
right procedure will be, it is really not surprising, because if you
look at the compared loudness curves test persons give, you'll see that
for low levels it's extremely "bath tub", whereas it flattens for more
volume. So the proposals speak of different filters for different volume
levels or even for variable filters etc. etc.

But I think just for a comparison of one design with another the unweighted
figures are just enough, it is only required that the very same method
is used for all test candidates.

But you're absolutely right in that you get "strange" or warped
values, no doubt.

Another observation: pulsed noise (clicks) were measured with
rms methods, i.e. some kind of mean value. Today experts seem
to agree that the peak value seems to be more meaningfull in
psychoacoustic terms.

This would be an argument for peak level detectors for envelope extraction.

m.c.