Hi, I have been recording some Christmas songs, and finally all seems
to be coming out well...but, I noticed that the peaks in the waveform,
recorded in Cool Edit Pro, seem to go up into the almost 30, 000 area.
There seems to be no clipping though. Do I need to squash these down a
bit then equalize and L1 them, or what is the best order to accomplish
this. What I am asking is, is that normal to have them going up that
high and low inside the track area, or should I be taking appropriate
action. See, I have been studying this compressing and equalizing
area, but can't get a grasp on how to even out these waveforms. Please
assist me so I can get these ready for CD burning, am sure you can
enlighten me on a simpler more tried and true method. Thanks so much,
for your input. Derrick

In article ,
(psongman) wrote:
Hi, I have been recording some Christmas songs, and finally all seems
to be coming out well...but, I noticed that the peaks in the waveform,
recorded in Cool Edit Pro, seem to go up into the almost 30, 000 area.
There seems to be no clipping though. Do I need to squash these down a
bit then equalize and L1 them, or what is the best order to accomplish
this.

If you want to trim down peaks, L1 is an appropriate tool. I don't know
what other tools you'd use before it that would be as transparent, so
I'd skip whatever else you had planned to put before it to handle peaks
and just use L1. L2 is probably a better choice, but L1 isn't bad as
long as you only use 3-5dB of gain reduction maximum, and make sure that
it's not doing gain reduction all that often.

Use your ears and listen to L1 when it's doing too much work so you
understand its artifacts. There's no reason why you can't get great
results with very few or no artifacts with a single pass of L1.

What I am asking is, is that normal to have them going up that
high and low inside the track area, or should I be taking appropriate
action.

When you record digitally, waveforms get stored pretty accurately, and
peaks like these will get stored and played back accurately. You can
just ignore this fact of nature and enjoy the accurate sound, but you'll
have a harder time making your mix sound extremely loud. If loud is
more important than good, then you need to shear off the peaks so you
can raise the average level without clipping. L1 or L2 is pretty good
at this, up to a point. I can't see how you'd need hyper loudness
beyond what L1 or L2 could do for a Christmas record though.

See, I have been studying this compressing and equalizing
area, but can't get a grasp on how to even out these waveforms.

Don't obsess about this. Go by what the mixes sound like, and skip the
waveform view. A single pass through L1 should be more than enough to
allow you to turn the level up to acceptable levels, especially for a
Christmas record.


Best of luck, and don't forget the 16 bit dither,

Monte McGuire

On 1 Nov 2004 20:08:09 -0800, (psongman) wrote:

Hi, I have been recording some Christmas songs, and finally all seems
to be coming out well...but, I noticed that the peaks in the waveform,
recorded in Cool Edit Pro, seem to go up into the almost 30, 000 area.
There seems to be no clipping though. Do I need to squash these down a
bit then equalize and L1 them, or what is the best order to accomplish
this. What I am asking is, is that normal to have them going up that
high and low inside the track area, or should I be taking appropriate
action. See, I have been studying this compressing and equalizing
area, but can't get a grasp on how to even out these waveforms. Please
assist me so I can get these ready for CD burning, am sure you can
enlighten me on a simpler more tried and true method. Thanks so much,
for your input. Derrick

As a part of my routine (I'm using the Sound Forge 6), I use the "Find
Largest Peak" tool, then reduce the peak found by -3 dB. To do this, I
stretch the waveform display just enough to allow me selecting the
peak so that the selection begins and ends in zero crossings. I do it
5 and more times a recording and upon normalization, I usually have a
dB or couple of them overall gain. But this is only an idea, to get
rid of excessive peaks without any limiting or compressing. There are
of course another solutions.

Edi Zubovic, Crikvenica, Croatia

Edi Zubovic wrote:
As a part of my routine (I'm using the Sound Forge 6), I use the "Find
Largest Peak" tool, then reduce the peak found by -3 dB. To do this, I
stretch the waveform display just enough to allow me selecting the
peak so that the selection begins and ends in zero crossings. I do it
5 and more times a recording and upon normalization, I usually have a
dB or couple of them overall gain...

....which is essentially a form of manual compression, which is great if
there are only a limited number of bad peaks (coughs, table knocks,
occasional record pops, etc.).

(psongman) wrote:

Hi, I have been recording some Christmas songs, and finally all seems
to be coming out well...but, I noticed that the peaks in the waveform,
recorded in Cool Edit Pro, seem to go up into the almost 30, 000 area.
There seems to be no clipping though.

CEP has a couple of tools to detect/handle clipping. When you play
the song the level indicator goes to over and that field stays lit
until you reset it by clicking on it. Another tool is "Analyze
statistics" which tells you how many samples are (possibly) clipped.
Having numbers in the 30,000 area is quite right. The maximum value
is 32767, the minimum -32768. When using 16 bits per sample CEP
considers an "over" when two or more adjacent samples are at the
limit.

Do I need to squash these down a
bit then equalize and L1 them, or what is the best order to accomplish
this.

If you are working in 16 bit mode you should care to not exceed the
range during EQ or any other operation. Most CEP function have a
level control to apply some attenuation to avoid clipping. Thus no
need change the levels manually, i.e. before EQ.

What I am asking is, is that normal to have them going up that
high and low inside the track area, or should I be taking appropriate
action.

If it sounds ok, you don't need to worry about compressing. However,
vocals sometimes sound a lot better when a light compressing is
applied to that track. That should be done before adding reverb.
From time to time I record a singer (soprano classical) that need
some light compressing. I use 5 ms look-ahead and 20 ms decay time
(very fast).

Usually I work in 32 bit (floating numbers) mode and don't care for
the levels at all until I convert to 44.1/16 for the CD. Depending
on the music genre I change the amplitude to + 0.5 ... + 2 dB in
32 bit mode, use the hard clip function in CEP to limit to -0.1 dB
(2 ms attack time, 7 ms decay time - very short again) and convert
the result to 44.1/16. Hard limit is always the last step before
changing the sample rate to 44.1/16 (CD).

Norbert