naturerecordists
[Top] [All Lists]

Re: Cool Edit Users.

Subject: Re: Cool Edit Users.
From: Walter Knapp <>
Date: Sat, 02 Nov 2002 22:38:28 -0500
Barb Beck wrote:

> There is a lot more to learn (most of it I do not know 8=3D))  but this s=
hould
> get you started.

I'm a lost sheep too. But here's a contribution. Sorry, you will have to
find where it is in cool edit elsewhere, I use macs, different software,
many of the same functions.

When working on basic cleaning up of a recording, I tend to think of
sound as a set of dimensions.

There is time, you have this nice recording, but right in the middle you
bump the mic. It's between two calls (if you are lucky). You can fix
this a couple ways. You can simply select the part of the timestream
containing the bump and delete it. If the calls you are recording are
not regular, and nothing else ended up with a jump in it's sound this is
fine. If not, you can go find a similar piece of sound elsewhere in the
recording and copy a piece of that, carefully noting the length of time
it occupies. Then replace the part with the bump by selecting and
deleting the exact same amount of time including it, and paste in your
clean piece. Again you have to be careful to check for breaks. Undo is
your friend. I'm not sure on cool edit, but if it can do these things
and crossfade the inserts or deletes this can help a lot. You can chop
out much larger chunks, of course. When I'm recording frogs, turning on
the recorder is turning on a car magnet. I may end up chopping out a
large chunk which is the cars running by. Or the dog that came silently
up behind me in the darkness one night and then barked from about 10'
away, a big dog. Wearing headphones that was amplified very nicely into
about a T-rex size dog.

Another is frequency. What you are specifically interested in will
almost certainly occupy only part of the frequency range. You can use
filters, lo cut, high cut, notch filters, equalizers and so on, all
remove or change part of the frequency band.  Note you can do very
little of this if interested in the entire soundscape, as it occupies
more of the frequency range. And if there is some call, even one you are
not interested in, that occupies a fairly wide range, you will find that
you have to remove it's full range. You will learn to like those things
that have a specific frequency they use thats narrow, and hate the wide
ranging callers.

Next dimension is sound energy. different parts of the sound are at
different dB levels, again some cover a wide range, some only a very
narrow range. There are filters that can selectively remove part of this
range. Dynamic range filters, and others. And many of the frequency
filters are settable as to how much they change the sound energy of what
they filter, allowing you to combine the filtering in two dimensions. A
straight dynamic range filter can be useful to remove low level noises
that cover wide frequency ranges. You might, for instance take
everything that's 60 dB below the peak sound level and cut it's energy
by 96 dB, effectively eliminating it. This one can be useful if you have
had to be severe in applying some other filters and they have produced
some low level artifacts. They can also be tricky, calls come wide or
narrow in this dimension too, amazing what damage removing the quiet few
percent of a loud call may do. Note on the frequency filters, most can
also increase the sound energy of the range they select, so you can go
either way.

You should become familiar with the sonogram facility in cool edit. Use
this as well as your ears to evaluate the effects of any filtering, etc.
you apply. The program I use on my mac for filtering I can set up a set
of various filters and insert the sonogram display anywhere in the
sequence to see what's happening. It will display while playing the
sound through the filter set. Not sure what all you can do with the one
in cool edit, but even if you have to run it separately, get in the
habit of checking the sonograms.

I believe cool edit has a noise filter. If it's the adaptive type, you
would have it sample a piece of your sound, then remove similar noise
from the rest. Sounds wonderful, but noise filters frequently produce
sounds of their own or distort sounds as the amount they are applied
increases. So you can generally only use them sparingly. Still, that can
be a big help.

Barb mentioned normalizing. This is a quickie gain adjustment where you
tell the program to apply as much gain as it can without clipping
anywhere in the recording. If your recording is fairly even with no
spikes this will bring the sound up quite dramatically sometimes,
sometimes too much. Other times there will be a few spikes, like the
bumping the dish above, and these will limit the gain. Remove the spikes
(assuming they are unwanted) and normalize again and you will get more
gain. Better if you can see the spikes before the first time remove them
  then. I use a somewhat more elaborate approach. My program will
analyze the soundfile and tell me just how much gain I can apply without
clipping. I normally don't apply all that, but depending on what I'm
doing I'll apply enough to give a set margin. For stuff I'm actively
editing, but need gain that might be to leave 2 dB margin. For other
uses it might be more. In the end, if it's all a bunch of tracks going
onto CD I'll adjust to a general level of loudness for the CD. This can
be more complex than just gain, as it's a function of things like
dynamic range, average and peak sound energy.

The order you apply various filters changes what you get too.
Experiment, but always have a raw, unfiltered original copy to go back
to if you mess up. You can spend a bunch of time on something, apply a
bunch of filtering, and somewhere it just falls apart, and the cause can
be anywhere along the chain or some combo. That's when you go back to
the beginning with a fresh copy and try again.

As Barb says, there's lots more. But there are a few thoughts to get
started. It takes time and lots of playing around with it to learn sound
processing. I don't think I'll ever get past feeling like a beginner on
that.

Walt




________________________________________________________________________
________________________________________________________________________



Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/


<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