This is correct(what Ted says). Some confusion can arise because
some people call the "bandwidth" of a filter
what is really one-half the bandwidth!
This is wrong but people sometimes do it
What Ted is giving you is the right thing.
Gerzon was not averse to making completely
unsubstantiated statements. I do not
think he had any real idea at all based on
solid evidence of
what kind of Q versus what kind of amplitude
error needed correcting. This is a complicated
thing --wide band errors(low Q) have lower
thresholds than narrow band errors(high Q),especially
on music material, and peaks
have lower thresholds than dips. What you need to make
things sound neutral is a complex issue.
Gerzon is pretending to information he did
not have here(mathematicians are prone to this--
they are so sure they are right! and they
usually are --about mathematics/ But psychoacoustics
is a hugely complex subject and knowing mathematics
does not make one know psychoacoustics. Improvisation is
the word for what is going on here. Gerzon,
for all the interest of his writings, is
full of this kind of stuff--made up and
he loves to claim authority. I did not know
him personally but he comes across as opinionated
and doctraire about complex issues, even though his stuff
makes interesting reading.) If you want to know
something closer to real answers about these questions
you could try Toole and associates--they too are opinionated
about what speakers ought to do(and in a rather
inane way) but their threshold work is ok I think.
--- In email@example.com, "Ted Rook" <rooknrol@...> wrote:
> A simple answer is "pretty sharp" ie, narrow bandwidth.
> Q is the center frequency (number of Hertz) divided by the width of the curve (number of
> Hertz) at the -3dB points either side of the center frequency.
> eg With curve bandwidth 1000Hz (at the -3dB points) and filter center frequency 1000Hz,
> This is for resonant circuits having symmetrical frequency response curves.
> On 22 Mar 2013 at 11:40, Alan Jordan wrote:
> > Hello,
> > The following quote is from one of the articles by Michael Gerzon
> > that
> > Robert Jorgensen linked to the other day:
> > "even the frequency response errors can only be corrected to a
> > limited
> > degree due to the one-third octave resolution of typical graphic
> > equalisers, which is coarser than the critical-bandwidth resolution
> > of
> > the human ear. Empirical experience of equalising PA systems with
> > parametric equalisers suggests that equalisers with a Q of about
> > 10
> > are required to tame simple frequency response effects well
> > subjectively."
> > My question is, what is a "Q of 10" in terms of octave width? Is
> > it
> > 1/10th of an octave?
> > Thanks,
> > Alan
> > ------------------------------------
> > Yahoo! Groups Links