naturerecordists
[Top] [All Lists]

Re: noise reduction-parametric eq-a basic example

Subject: Re: noise reduction-parametric eq-a basic example
From: Walter Knapp <>
Date: Wed, 01 Dec 2004 19:50:23 -0500
From: Marty Michener <>

> In practice, the continuous curve which COOL lets you manipulate at each=

> point apparently proves to be too much intellectually for most nature sou=
nd
> users, anyway, judging from the total lack of follow-up on these
> discussions, so your parametric equalizer probably works better for most=

> users.  And yes, I do seem to be making a very snobby point, here, so let=
's
> please hear some responses.  [Any response would be better than those I
> have received on the last three questions I have asked. . . .=3D none wha=
tever.]

My past system was primarily based on appropriate individual filters set
at appropriate frequency points, using a stack of them in a filter
matrix. This would be followed in the stack by a trainable noise filter
applied sparingly. And then at the end of the stack a realtime sonogram
to see what happened.

I'm slowly moving to Elemental Audio's Equium and Firium plug in's to
cover a lot of the equalizer functions:
http://www.elementalaudio.com/products/index.html
These provide quite a bit more bells and whistles than a straight
parametric. Creating a whole batch of frequency based filtering that can
be used at once. Has a nice graphic interface.

I recently got their new plug-in Neodynium, I'm hoping that will replace
the simpler dynamics processor in SoundHack that I've been using. In
Elemental Audio's own words Neodynium is a graphic, multi-regional
dynamics processor. Fundamentally it allows you to apply compression in
bands of audio levels. The SoundHack dynamics processor I would use to
drop very quiet filtering artifacts down out of audibility, this looks
like a more advanced version of the same thing. Dynamics processing is
tricky in any case.

> I just got back from a week's (computer) absence, more or less.  The answ=
er
> to your earlier question about COOL: "Why not be able to decrease the
> sounds at ZERO frequency", is --- there really IS no zero frequency.
> Logically, this would be DC =3D direct current, with no variations in vol=
tage
> ever - at least for a very long time -- several millennia?!  So the COOL=

> software realistically sets the lowest frequency it can reasonably contro=
l
> to 20 Hz -- mas o menos.  Practically, 20 - 30 Hz means zero to you and
> me.  Not to worry.

That is unless you possess and record with a MKH-110. Then 20 is only
part way there. Infrasound, if you can record it and play it back, has
considerable effect on the listener. A lot of the emotional content of
sound is down there.

> My oft-repeated rule I hereby reiterate and develop: The final files soun=
d
> best, at least for bird recordings to most laymen, (I don't know about
> frogs or loud roaring beasties files) when the lower frequency realm
> closely approximates equal sound energies for every octave, with no singl=
e
> dominating tone or set of overtones.  If you do have a loud hum, then
> reduce it to EQUAL the other sounds, on average.  But if you leave a file=

> with missing frequency ranges between 20 and 2000 Hz, laymen will notice =
a
> "hollow" sound, which often seems to bother normal listening.

I would not always go along with this, though it's probably about right
for listening, ambient recordings. A light hand is always best here.

But, if your purpose is to hack out a sample clip of a single species
call from the whole you can get much more elaborate with the filtering.

Walt





________________________________________________________________________
________________________________________________________________________

<Prev in Thread] Current Thread [Next in Thread>
  • Re: noise reduction-parametric eq-a basic example, Walter Knapp <=
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