naturerecordists
[Top] [All Lists]

Re: effect of mic sensitivity

Subject: Re: effect of mic sensitivity
From: "Eric Benjamin" ericbenjamin2
Date: Sun Feb 24, 2008 5:36 pm ((PST))
> I note your careful wording or caveats
Yes indeed, I did carefully leave some outs so that no one can blame me for=
 what I said!  That is, don't shoot the messenger.
but...

> I have always had a huge problem in believing in Sones
> Seems to me to be "dodgy science"
> How the F*** can anyone say sound x is twice as loud as sound y?
> the equivalent of how orangey is fruit A than fruit B

You're not wrong to express these doubts.  At the very least one will have =
the questions of how difficult it is to make these determinations, how repe=
atable they are, and also how useful they are.

The answer to the third question is that objectifying subjective qualities =
can be very useful.  As an example, let us suppose that I am in the busines=
s of selling oranges, and I wish to grade them as to their beauty, then it =
is to my advantage to be able to measure qualities like "orange-ness".  OK.=
  We in this group don't care about how orange things are.  But suppose you=
 ask me to "Make it twice as loud", but I only have a knob or a slider that=
's labeled in dB.  Then I need to have at least some sort of a guide as to =
what 'twice as loud' is, in dB.

Part of the utility of this sort of thing has to come from the (desired) fa=
ct that different person's perceptions are similar.  I haven't done these p=
articular experiments (loudness growth), but I have done experiments with l=
oudness matching, and some pretty remarkable things came out of that.  I as=
ked my subjects to match the loudness (by changing the level) of varying au=
dio recordings.  Something like making the loudness of a car crash equivale=
nt to the loudness of an operatic soprano.  A typical subject reaction is "=
I can't do that.  It's like comparing apples and oranges."  But it turns ou=
t that they then proceed to do it, and that expert subjects (mixing enginee=
rs and such like) can match loudness levels repeatably to within an increme=
nt as small as 1/4 dB.  Which is quite remarkable.

Now, it does turn out that different person's judgments do differ, but not =
as much as you might think.  If you take a large collection of recordings o=
f frog vocalizations, and adjust them so that YOU think that they are all t=
he same loudness, then it turns out that you will almost certainly (95% con=
fidence) be within a couple of dB of what other people think is equal loudn=
ess.

Of course, you may not want the frog recordings to be equal loudness...

We're on the edge of something that is purely personal (like, what does the=
 color 'red' mean to you?), but that is far too difficult for me to tackle.=
  What is of meaning to me is that if someone asks me to make it twice as l=
oud, I can turn up the level by 10 dB and be close to what it is that they =
want.

Eric






<Prev in Thread] Current Thread [Next in Thread>
Admin

The University of NSW School of Computer and Engineering takes no responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the naturerecordists mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU