[Keith W. Blackwell]

Jacques E. Bouchard wrote:
We'll be dubbing this week, and I'm either going to find a dead room or a
very quiet outdoor location. The latter would allow me to reproduce the
acoustics, but there may be uncontrollable factors (I tried one location
and could not identify an annoying rumble until I spotted a highway about
a mile away).

I think you said you were using Adobe Audition. Pardon me
if I misread that.

I'm very familiar with Cool Edit Pro, so Audition should be
similar. If you can get a fairly dead room, you may still have
some boom to eliminate; if nothing else is available, try a
walk-in closet with lots and lots of clothes, then cover the
remaining surfaces with blankets. Ok, I'm letting my low-
budget focus show here.

In post, I've had limited success with using the Dynamics
effect to expand downward the lower-level content. You can
limit the frequency range to, say, 20 -- 200 Hz. Use a curve
that is linear diagonal except for a spike where you are
considering putting your drop -- this will accentuate the
particular sounds that will be affected, allowing you to
narrow in on an effective position (though it is difficult
to adjust, since a spike will take a few points in the curve).
Select a short but representative section of your track to
preview the effect on as you adjust the parameters.

You want to find the low-frequency reverb at the end of
uttered syllables. Ideally, the effect will not have any
work to do during continuous utterences.

You'll also need to adjust the attack
and release times in the detector and actuator so that it
works fast enough to catch the trail-off but resumes slowly
enough to have a natural sounding recovery. Once you've
isolated the right level to place this adjustment, set that
point to drop fairly steeply (downward expansion) and to
approach the bottom left according to whatever curve you
favor.

It's totally different from any other dynamics processor
interface out there, but it is mathematically reasonable
and if you take the time to understand it -- it can be a
very useful tool. Unfortunately, this can only remove the
LF reverb between and after syllables, but you'll still
hear some crossing over when there is no gap. So it's not
perfect, but it can help eliminate some of the more
noticeable "boom" from your room sound.

There are other tricks to try with it. You can also try
reversing the wav and applying it in reverse -- with
slower attack/release times, this will allow you to keep
LF content only in the middle of the syllables where they
are loudest. You can also introduce compression above
that point to simply reduce the dynamic range of the LF
content from there up.

You're just *so* much better off using a dead room to
start with. If your talent can keep a steady distance
and avoid popping, then you can use proximity effect of
a directional mic to allow you drastically EQ the lows
back down just to get it to sound normal again, which in
turn helps you get a better direct-to-reflected ratio in
the lower frequencies. And of course it takes a lot of
roll-off.


But you probably already know all that. I mainly just wanted
to mention the band-limited downward expansion thing and
another thing...

Another CEP weirdity that you might still have in Audition
is that to use Convolution you have to first have a separate
wave file containing the impulse response you want to convolve with.
I would assume you would want to use this to introduce some
simulated ambience to match the supposed setting, while still
recording everything in your "dead room".

Keep it to a reasonable length, like less than half a second.
You "copy" the wave, then go back to editing the one you want
to convolve with it. There, you bring up the Convolution
window, clear it all out, and you can "paste" your wave into
it. CEP did store these convolution impulses in its own
proprietary format so that it could keep track of scaling and
so on (you get to "Save" your impulse). Note that if you
want true stereo, you will need a stereo impulse for the left
position and another stereo impulse for the right position. Clone
your original track into two, one to have the original Left
in both channels, the other to have the original Right in both
channels. Apply corresponding impulses to each, then remix
the two back together. Pain the the butt, but it works.
In other words, you have to have 4 channels to convolve 2
into 2 properly. But you probably don't care about true stereo.
Audition may make that easier. I dunno. Once you've got the
hang of it, it is quite easy to record your own impulses by
popping a balloon, using a cap gun, a spark plug rig, or
whatever.


--
Keith W. Blackwell
(I do not speak for my employer or anyone else)