Normalizing vs. Not normalizing?
MANDRAKE
I remember craig anderton advising against this in a magazine, but
can't remember why. Anybody does?
Cheers,
/R
Sent via Deja.com http://www.deja.com/
Before you buy.
plus4dbu
MANDRAKE <rbc...@my-deja.com> wrote:
> Shoud one normalize all his samples (peak value)?
>
> I remember craig anderton advising against this in a magazine, but
> can't remember why. Anybody does?
The reason why is because it rarely does any good. What normalizing
does is finds the peak level in the section you're normalizing, and
amplifies the whole thing by just enough to bring that peak up to full
level (0 dBFS).
If the overall record level was low, say the highest peak was -10 dBFS,
then normalizing will bring everything up by 10 dB, including the noise.
It will make it louder, but it will also make it noisier. You can
always do this manually in the mix, so there's no t much point in
letting the computer muck with it. You really should have recorded at a
higher level, and hopefully you will, next time.
If, on the other hand, the level was mostly around -10 dBFS, but there
was just one spike that went all the way up to, say, -1 dBFS, then
normalizing would raise the level only 1 dB, which won't make it sound
any louder.
There are some programs now which have a smarter normalizing function,
more like compression than actually normalizing. It will determine the
average level (RMS or something else) of the section you're normalizing,
will raise that, and sit on the peaks to keep them from clipping. This
may be more useful.
In any case, give it a try. That's why these things have an Undo button
on them.
Sean & Jenelle Wright
plus4dbu wrote:
i prefer to normalize, then remove silence, then compress (if necessary).
since i usually compress during tracking, the normalizing pushes the average
level up. if your recording isn't particularly hot, then removing silent &
not quite so silent passages can cause you to lose your trails during
digital processing.
izzy
Jay - Atlanta Digital
> Shoud one normalize all his samples (peak value)?
>
> I remember craig anderton advising against this in a magazine, but
> can't remember why. Anybody does?
Normalizing has a place, but in general it's bad more often than good in
common practice. Every DSP operation can potentially add noise and
degrade audio quality, and add truncation distortion if not dithered.
Also, successive DSP processes really can degrade sound quality. On the
other hand, if you are doing a small gain change and dithering correctly,
and it's the last step of the process (besides dithering of course), that
is, not going to mastering, not having any other processing done, not had
a gain change done already, then it can be OK.
-Jay Frigoletto
(Atlanta Digital)
We've moved to Hollywood!
http://www.promastering.com
Keith Sklower
Jay - Atlanta Digital <atl...@aol.com> wrote:
}In article <84a96q$qa7$1...@nnrp1.deja.com>, MANDRAKE <rbc...@my-deja.com> wrote:
}> Shoud one normalize all his samples (peak value)?
}> I remember craig anderton advising against this in a magazine, but
}> can't remember why. Anybody does?
} ... if you are doing a small gain change and dithering correctly,
}and it's the last step of the process (besides dithering of course), that
}is, not going to mastering, not having any other processing done, not had
}a gain change done already, then it can be OK.
There has been some interesting discussion on the pro-audio mailing list
lately about the possiblity that if you have normalized your material, overflow
(clipping) might occur in the antialiasing filters during D/A if they
those filters are not properly implemented.
On the other hand, if I'm making a CD/R dup for a member of the community
orchestra or band that I play in of one of our concerts, I figure that
odds are that they aren't going to want to bother about having to walk
to their stereo system and crank up the sound level, so I go ahead an
normalize it anyway....
MANDRAKE
> If, on the other hand, the level was mostly around -10 dBFS, but there
> was just one spike that went all the way up to, say, -1 dBFS, then
> normalizing would raise the level only 1 dB, which won't make it sound
> any louder.
>
remove the peaks manually... ooops..
Cheers,
/R
Jay - Atlanta Digital
> > If, on the other hand, the level was mostly around -10 dBFS, but there
> > was just one spike that went all the way up to, say, -1 dBFS, then
> > normalizing would raise the level only 1 dB, which won't make it sound
> > any louder.
> >
> I know you won't like this... but personally I use the pencil tool and
> remove the peaks manually... ooops..
>
The pencil tool is bad! It's sometimes necessary if you have nothing else,
and it gets you out of certain jams, but not in a way that you could call
"high quality". A person cannot manually draw a complex non repeating
waveform by hand and expect the correct spectral content to match. Then
again, the flawed result is often better than the larger original flaw.
Something like Sonic Solutions' Manual declicking is a better answer. It
analyzes the area just to the left and just to the right (and there are
some other variables you can address too in case you're right next to a
cymbal or something) and interpolates what was likely to have been there
had the problem not occured. In other words, it synthesizes a piece of
audio with the proper spectral content and amplitude and plugs it into the
problem spot. Or in this case of normalizing, one could isolate the
spike, and instead of drawing it away, you could just reduce it a few db
and give yourself room for the gain increase you desire. Not for every
occasion, but another option to look at in this situation. Use your ears
to see which option at your disposal sounds best.
Jay - Atlanta Digital
blin...@aol.com (Blind Joni) wrote:
> > A person cannot manually draw a complex non repeating
> >waveform by hand and expect the correct spectral content to match.
>
> Just for the info..I use a PC with Samplitude and one of the pencil options is
> a volume redraw that keeps the waveform of the peak and redrws it at a lower
> volume. I have found this invaluable for fixing a lot of things such a vocal
> consonants that are too loud or percusion hits that the attack is a little to
> sharp. I don't know if other programs offer this but I fnd it much better than
> trying to draw a waveform myself.
>
> John A. Chiara
That sounds like a good feature. I haven't seen that anywhere else
myself, but it's like the other suggestion in my post. I often just make
a little segment of it and attenuate. I only pull out interpolation tools
if it's something ugly, not just a peak that would be OK if it weren't so
big. Sound like the samplitude programmers did that too, and decided to
make an easy tool for it. Much better in that type of situation than
drawing a waveform with a pencil.
Blind Joni
>waveform by hand and expect the correct spectral content to match.
Just for the info..I use a PC with Samplitude and one of the pencil options is
a volume redraw that keeps the waveform of the peak and redrws it at a lower
volume. I have found this invaluable for fixing a lot of things such a vocal
consonants that are too loud or percusion hits that the attack is a little to
sharp. I don't know if other programs offer this but I fnd it much better than
trying to draw a waveform myself.
John A. Chiara
SOS Recording Studio
Albany, NY
"Survivor of the Slums"
Mike Rivers
> Just for the info..I use a PC with Samplitude and one of the pencil options is
> a volume redraw that keeps the waveform of the peak and redrws it at a lower
> volume.
Could you get to the same place simply by highlighting the peak and
applying a volume reduction to the highlighted area? It still
wouldn't sound the same as the original, but it would preserve the
waveform pretty much.
I've never been able to draw anything recognizable with a mouse in a
paint program. Why they think I could redraw a sound-alike waveform
I'll never know.
--
I'm really Mike Rivers (mri...@d-and-d.com)
Jay - Atlanta Digital
> In article <19991230215658...@ng-ch1.aol.com>
blin...@aol.com writes:
>
> > Just for the info..I use a PC with Samplitude and one of the pencil
options is
> > a volume redraw that keeps the waveform of the peak and redrws it at a lower
> > volume.
>
> Could you get to the same place simply by highlighting the peak and
> applying a volume reduction to the highlighted area? It still
> wouldn't sound the same as the original, but it would preserve the
> waveform pretty much.
Mike, haven't you been paying attention? It's already been mentioned that
you can make a little segment there and attenuate. If you attenuate a
small enough area, and not too aggressively, it really flows by
seamlessly. Sometimes I'll grab a little spot of sibilance in this same
way. If there's just a couple in a track, it makes no sense to globally
apply a de-esser. Of course, depending on what else is under there, this
may or may not be possible.
Jon Best
mix, and sometimes that works better than a straight volume change. No idea why.
Jon Best
Sales Weasel From Mars
Robert Herman
Keith Sklower wrote:
>
> In article <atldigi-2812...@user-2iniuf8.dialup.mindspring.com>,
> Jay - Atlanta Digital <atl...@aol.com> wrote:
> }In article <84a96q$qa7$1...@nnrp1.deja.com>, MANDRAKE <rbc...@my-deja.com> wrote:
>
> }> Shoud one normalize all his samples (peak value)?
>
Depends on what you want---
If you want the maximum sample value in the file to be 100% then
Normalize- This is typically a good idea because you are using more bits
to represent the sample data especially if more processing is to be
done.
If you don't want to affect the overall volume level, then don't.
If you want to maximize the overall volume level and compress the
dynamic range. Use Waves UltraMaximizer
Rob Herman
Stone Studio Productions
Chicago
Jim Gilliland
>
> In article <19991230215658...@ng-ch1.aol.com> blin...@aol.com writes:
>
> > Just for the info..I use a PC with Samplitude and one of the pencil options is
> > a volume redraw that keeps the waveform of the peak and redrws it at a lower
> > volume.
>
> Could you get to the same place simply by highlighting the peak and
> applying a volume reduction to the highlighted area? It still
> wouldn't sound the same as the original, but it would preserve the
> waveform pretty much.
Basically, this process is just acting as a sort of manually applied
limiter, so it makes some sense to use your software's
compression/dymamics routines to adjust these levels. By selecting your
boundaries carefully and using both an aggressive attack and release
time, you can effectively reduce the volume without much damage to the
waveform, and without an abrupt transition at the beginning or end of
the section. Works well for short duration peaks (say, less than a
second).
Mike Rivers
> > Could you get to the same place simply by highlighting the peak and
> > applying a volume reduction to the highlighted area? It still
> > wouldn't sound the same as the original, but it would preserve the
> > waveform pretty much.
>
> Basically, this process is just acting as a sort of manually applied
> limiter, so it makes some sense to use your software's
> compression/dymamics routines to adjust these levels. By selecting your
> boundaries carefully and using both an aggressive attack and release
> time, you can effectively reduce the volume without much damage to the
> waveform, and without an abrupt transition at the beginning or end of
> the section. Works well for short duration peaks (say, less than a
> second).
I was talking about much less than a second, like a few milliseconds.
That's a peak. Something that's half a second long and clipping all
the time is just a fuckup when it came to setting level. I wouldn't
expect to be able to make that sounds decent since it would be
composed of many cycles all clipping.
The initial discussion was using a "pencil tool" to redraw the
flattened top of a cycle, I believe. The bottom line is that you just
shouldn't let this sort of thing happen. If it does, there are some
things that can make it sound less horrible, but still not sound
right.
Jim Gilliland
>
> In article <386F4E2D...@altavista.net> usemyl...@altavista.net writes:
>
> > > Could you get to the same place simply by highlighting the peak and
> > > applying a volume reduction to the highlighted area? It still
> > > wouldn't sound the same as the original, but it would preserve the
> > > waveform pretty much.
> >
> > Basically, this process is just acting as a sort of manually applied
> > limiter, so it makes some sense to use your software's
> > compression/dymamics routines to adjust these levels. By selecting your
> > boundaries carefully and using both an aggressive attack and release
> > time, you can effectively reduce the volume without much damage to the
> > waveform, and without an abrupt transition at the beginning or end of
> > the section. Works well for short duration peaks (say, less than a
> > second).
>
> I was talking about much less than a second, like a few milliseconds.
> That's a peak. Something that's half a second long and clipping all
> the time is just a fuckup when it came to setting level. I wouldn't
> expect to be able to make that sounds decent since it would be
> composed of many cycles all clipping.
I'm not talking about something that clipped - that would need a whole
different kind of repair. I'm talking about a peak - where there is a
sudden short-duration jump in level. It could be a few milliseconds, or
it could be 500. Either way, this approach can work.
Of course, if it's short enough, and if it didn't belong in the signal
to begin with, you can just delete it or silence it.
Keep in mind that my recordings are made in live stage situations, so I
wind up with lots of little things like this that need to be fixed:
sibilance, popped "P"s, bumped mics, etc. I don't have the luxury of
getting another take, using a studio pop filter, or teaching mic
technique. So I sometimes have to fix it after the fact.
MANDRAKE
> >waveform by hand and expect the correct spectral content to match.
>
options is
> a volume redraw that keeps the waveform of the peak and redrws it at
a lower
a vocal
> consonants that are too loud or percusion hits that the attack is a
little to
> sharp.
That's actually really cool,
I'll try that, however my real problem with samplitude is the user
interface and the weird mouse usage...
Anyway, I don't see the 'pencil' as evil, it is very very handy... (As
a matter of fact I prefer cooledit where you can move the single
offending sample), for clicks and discontinuities I found the pencil to
behave better than most of the 'click removal' software.
R.
www.digido.com It's not how loud you make it but how to make it loud.
Hope this helps,
R.
On Tue, 28 Dec 1999 12:14:59 GMT, MANDRAKE <rbc...@my-deja.com> wrote:
>Shoud one normalize all his samples (peak value)?
>
>I remember craig anderton advising against this in a magazine, but
>can't remember why. Anybody does?
>