For me, its useful to always remind myself that its not really
possible to "remove" sounds. I conceptualize "post" processes as
adjustments to the overall tonal balance and dynamics so that the
experience, upon playback, is more like the one I had in the field.
There are several technical and site-dependent factors that make
sound playback experience un-life-like. I usually have to address all
of these influences at once in post so thinking only in terms of
"noise" is usually not comprehensive enough.
As long as I use very low-noise mics and mic preamps, I no longer get
huge, objectionable amounts of recording system "noise" that I can't
address effectively with parametric EQ. If this type of noise is the
problem, the best solution is in hardware.
Not to say that some "noise" reduction apps don't work better with
multiple, "light" passes, but if it was my job or passion to design,
test or write code for a "noise" reduction application, chances are
good that I'd consider the possibilities of using multiple passes. If
I found that multiple passes improved the performance of my software,
I would try including this feature in the application.
I've come across same, basic observations that Scott has. The "noise"
reduction applications I've used (DINR, Sound Soap, those built into
Logic and a half dozen others) tend to be more effective on
sinusoidal-like sounds than other sound types. (Pure, sinusoidal
sounds are fixed in pitch and amplitude). Even so, when I resort to
a "noise" reduction plug, it has often helped to insert a parametric
EQ plug before the "noise" reduction plug to reduce the presence of
the bands that the noise reduction plug is having the most difficulty
addressing. If the "noise" varies in amplitude, it can also help to
insert compression between the parametric EQ plug and the noise
removal plug. These are not easy adjustments to make. The "noise
reduction" apps tend to address HF elements better. Note that
de-essing plugs can also work fairly well for some forms of
fluctuating HF "noise." For LF shaping, plugs like Neodynium combine
frequency and compression.
If the multiple pass approach works, an A/B test might reveal
something,.. Rob D.
=3D =3D =3D
At 8:51 AM -0500 6/23/08, Curt Olson wrote:
>Rob makes fine recommendations here for standard processing. But for
>digital noise reduction, multiple passes of lighter processing is
>indeed best.
>
>Curt Olson
>
>On Jun 23, 2008, at 6:22 AM,
><naturerecordists%40yahoogroups.com>=
m
>wrote:
>
>> 1e. Re: Filtering
>> Posted by: "Rob Danielson" <type%40uwm.edu>
>>danielson_audio
>> Date: Sun Jun 22, 2008 6:52 pm ((PDT))
>>
>> --- In
>><naturerecordists%40yahoogroups.com>=
om,
>>"Chris Hails" <>
>> wrote:
>>>
>>> --- In
>>><naturerecordists%40yahoogroups.com>=
com,
>>>"Aaron Ximm" <aaron.ximm@>
>>> wrote:
>>>>
>>>> Someone here has said on occasion that it is better to do a few
>> very
>>>> light passes, than one deep passage.... was that you Dan? Rob?
>>>>
>>>> best,
>>>> aaron
>>
>> Several years back, I described using "stacked plugs" as a work-
>> around for parametric EQ plug-ins that only provide a few bands when
>> more bands are required. I can't think of a reason to use mulitple
>> passes when one can makes all the necessary changes at once. I try
>> to follow the "simpler is better" axiom and often combine several
>> types of digital processing like EQ, balance, gain, phase and
>> increasing bit depth into one, final, step when out-putting a file.
>> When I'm having trouble judging which equalization settings/effects
>> produce the best results, I output the varied attempts and compare
>> them back to back in a timeline.
>>
>> I can see the merit in taking time with the judgements, but not in
>> spreading the decisions out over multiple processing passes. Rob D.
>
>
--
|