Re: Normalization Pros and Cons?

[Steven Sena]

Wow....You sound like you have been hanging round some of my clients...
Mr. Dorsey's advice is excellent.

In my own words it would sound more like (in my best southern draw)...Hey,
take it easy man...It's just rock and roll...

--
Steven Sena
XS Sound
www.xssound.com


"Alan Cassaro" > wrote in message
...
> I have quite a few finished mixes from my Cakewalk projects, and I'm
> just about ready to do my final mastering. This involves hundreds of
> master mixes and songs, as well as multiple mixes of many of the same
> songs. I had been under the impression, perhaps mistakenly, that
> normalization was basically the same thing as volume gain. In order for
> me to make good sound comparisons of my work, I normalized most of the
> songs as I went along, mostly to 90 percent, rather than a hundred
> percent. However, I have a lot of OTHER tracks from old tape masters
> that were digitally transferred over to DAT tape by a respected
> engineer, who recorded them at such a low level that most of them were
> about 50 percent of full gain. He said, "No problem, you can raise them
> later in the PC, it's all digital, and it doesn't matter, you won't be
> adding any hiss." He wanted to allow himself plenty of headroom, he
> said. So, I went ahead and raised them in the computer by normalizing,
> rather than by using my volume gain in my sound editor. (It's easier to
> do these by batching them all at once, rather than by doing them one at
> a time."
>
> My personal theory was that I could do further processing (eq, effects,
> compression, etc) on the songs because I had left myself plenty of
> headroom. Was I wrong to make this assumption?
>
> In the past several weeks I have been told by another engineer that I
> have to start from scratch and create new mixes because the
> normalization I have added has probably already colored my sound and
> ruined the dynamics. He said that I should ALWAYS use the volume gain on
> my sound to approximate 90 percent, rather than normalize to 90
> percent.
>
> However, he could NOT explain satisfactorily to me just what it is that
> normalization does to "ruin" the dynamics of my songs, only that "he"
> can hear the difference. Personally, I can't, but I am so close to the
> project and material that I freely admit I have lost a lot of my
> objectivity.
>
> I dread the thought of having to go back to my masters to "start from
> scratch"; it's giving me nightmares.
> Questions:
> 1. Is it okay to do further processing on the tracks that I normalized
> to 90 percent?
> 2. Can I lower the gain on the songs that may have been normalized to
> one hundred percent, in order to do further processing.
> 3. Do I have a different problem with the songs that were transferred
> digitally at only 50 percent of full volume, and then normalized later?
> 4. Some songs have been normalized more than once, but always under a
> hundred percent. Does this create problems?
>
> 5. Please explain how and why normalization "ruins" a songs dynamics and
> coloration, versus using a gain control to achieve the same approximate
> sound level, particularly since I am working much lower than full
> volume. After all, if normalization finds the loudest peak, isn't it
> just acting as a gain control ?
>
> Thanks in advance for your response. Please don't beat me up on this, I
> feel pretty bad about this, because of conflicting opinions I have
> gotten from two different engineers.
>
> By the way, most of the music I am working with is basic guitar driven
> rock and roll (rockabilly, CCR styled), some pop. The music is all
> pretty dynamic, compressed and loud, it's certainly isn't chamber music,
> or music with any quiet passages.
> Alan
>
>
>

[Ben Bradley]

In rec.audio.pro, Alan Cassaro >
wrote:

> ...

>My personal theory was that I could do further processing (eq, effects,
>compression, etc) on the songs because I had left myself plenty of
>headroom. Was I wrong to make this assumption?

It depends on the processing. If it's compression or limiting
(which lowers the level, unless you turn up the makeup gain too much),
there should be no problem. EQ could be a problem and cause clipping,
especially if you boost a frequency range. Even cutting a range could
cause clipping - a high-Q filter hit by a transient could ring like a
bell.
Can you tell when something is clipping? I (usually) can. Just try
whatever it is you want to do and and see (or hear).

> In the past several weeks I have been told by another engineer that I
>have to start from scratch and create new mixes because the
>normalization I have added has probably already colored my sound and
>ruined the dynamics. He said that I should ALWAYS use the volume gain on
>my sound to approximate 90 percent, rather than normalize to 90
>percent.

Try a listening test something like I recall Arny described before.
Take a recording and copy it to a scratch track. Reduce it by 10dB,
then raise it back up 10dB. Compare this to the original and see if
you can hear any difference. Do likewise with 20, 30, 40, 50, 60, and
70dB. Somewhere you should start hearing the noise. If things are not
dithered properly in the gain changes, it should also start to sound
'gritty' or distorted.
Keep in mind that the 'normalized to 50 percent' you talk about
below is only 6dB below digital clipping.

>However, he could NOT explain satisfactorily to me just what it is that
>normalization does to "ruin" the dynamics of my songs, only that "he"
>can hear the difference. Personally, I can't, but I am so close to the
>project and material that I freely admit I have lost a lot of my
>objectivity.

Maybe try the above listening test with a ripped CD track or
something else you can be 'objective' about.

>I dread the thought of having to go back to my masters to "start from
>scratch"; it's giving me nightmares.
>Questions:
>1. Is it okay to do further processing on the tracks that I normalized
>to 90 percent?
>2. Can I lower the gain on the songs that may have been normalized to
>one hundred percent, in order to do further processing.
>3. Do I have a different problem with the songs that were transferred
>digitally at only 50 percent of full volume, and then normalized later?
>4. Some songs have been normalized more than once, but always under a
>hundred percent. Does this create problems?

Is there some option to do dithering when changing gain or
normalizing? If so, it should always be on.

>5. Please explain how and why normalization "ruins" a songs dynamics and
>coloration, versus using a gain control to achieve the same approximate
>sound level,

Is the 'gain control' between the computer and your ears, or is it
a slider that affects the level being put into another .wav file? If
the latter, then it's changing the gain the same way normalizing does.

>particularly since I am working much lower than full
>volume. After all, if normalization finds the loudest peak, isn't it
>just acting as a gain control ?

Yes.

>Thanks in advance for your response. Please don’t beat me up on this, I
>feel pretty bad about this, because of conflicting opinions I have
>gotten from two different engineers.

Everyone has opinions, and you know what opinions are like...

>By the way, most of the music I am working with is basic guitar driven
>rock and roll (rockabilly, CCR styled), some pop. The music is all
>pretty dynamic, compressed and loud, it's certainly isn't chamber music,
>or music with any quiet passages.

I like Scott's advice. Have some tea, and maybe even listen to some
music with quiet passages. Try Beethoven's 6th.

>Alan

[Roger W. Norman]

Hard to beat Scott's answers so I won't try. I'll just suggest that you
don't touch the original files, either by making copies, which can become a
little real estate expensive with a hard drive, or by using non-destructive
editing processes so you can back out or start over. In the applications
I've worked with, normalization is a destructive process, meaning it changes
the original file unless you've remembered to rename it. Just leave the
original alone and make all your adjustments within the mixing environment,
including bringing up the volume in the last stage of the mixing process.
That way you can scrap the edl and redo without having destroyed your
original file.

Other than that, Scott's answers are right on. Don't do multiple operations
when you can do one. If you've EQ'd a file you don't want to go back and re
EQ it, you want to undo the EQ and then EQ again. This way you don't get
cumlative errors, one based upon the other. And if you've got the mixing
horsepower, don't write any volume changes or EQ, dynamics processing or
whatever, to the file (this is a function that some use to take individual
files, say a guitar track, and bounce it down to a new EQ'd or whatever
track to help keep processing requirements down). If you start running out
of horsepower, start using aux sends for some processing so that you
eliminate some of the instantiations of processes. Why use 5 reverbs, for
instance, when you can run 5 tracks through one instance of a reverb on an
aux send? 1/5 the processing power necessary.

--


Roger W. Norman
SirMusic Studio
Purchase your copy of the Fifth of RAP CD set at www.recaudiopro.net.
See how far $20 really goes.




"Alan Cassaro" > wrote in message
...
> I have quite a few finished mixes from my Cakewalk projects, and I'm
> just about ready to do my final mastering. This involves hundreds of
> master mixes and songs, as well as multiple mixes of many of the same
> songs. I had been under the impression, perhaps mistakenly, that
> normalization was basically the same thing as volume gain. In order for
> me to make good sound comparisons of my work, I normalized most of the
> songs as I went along, mostly to 90 percent, rather than a hundred
> percent. However, I have a lot of OTHER tracks from old tape masters
> that were digitally transferred over to DAT tape by a respected
> engineer, who recorded them at such a low level that most of them were
> about 50 percent of full gain. He said, "No problem, you can raise them
> later in the PC, it's all digital, and it doesn't matter, you won't be
> adding any hiss." He wanted to allow himself plenty of headroom, he
> said. So, I went ahead and raised them in the computer by normalizing,
> rather than by using my volume gain in my sound editor. (It's easier to
> do these by batching them all at once, rather than by doing them one at
> a time."
>
> My personal theory was that I could do further processing (eq, effects,
> compression, etc) on the songs because I had left myself plenty of
> headroom. Was I wrong to make this assumption?
>
> In the past several weeks I have been told by another engineer that I
> have to start from scratch and create new mixes because the
> normalization I have added has probably already colored my sound and
> ruined the dynamics. He said that I should ALWAYS use the volume gain on
> my sound to approximate 90 percent, rather than normalize to 90
> percent.
>
> However, he could NOT explain satisfactorily to me just what it is that
> normalization does to "ruin" the dynamics of my songs, only that "he"
> can hear the difference. Personally, I can't, but I am so close to the
> project and material that I freely admit I have lost a lot of my
> objectivity.
>
> I dread the thought of having to go back to my masters to "start from
> scratch"; it's giving me nightmares.
> Questions:
> 1. Is it okay to do further processing on the tracks that I normalized
> to 90 percent?
> 2. Can I lower the gain on the songs that may have been normalized to
> one hundred percent, in order to do further processing.
> 3. Do I have a different problem with the songs that were transferred
> digitally at only 50 percent of full volume, and then normalized later?
> 4. Some songs have been normalized more than once, but always under a
> hundred percent. Does this create problems?
>
> 5. Please explain how and why normalization "ruins" a songs dynamics and
> coloration, versus using a gain control to achieve the same approximate
> sound level, particularly since I am working much lower than full
> volume. After all, if normalization finds the loudest peak, isn't it
> just acting as a gain control ?
>
> Thanks in advance for your response. Please don't beat me up on this, I
> feel pretty bad about this, because of conflicting opinions I have
> gotten from two different engineers.
>
> By the way, most of the music I am working with is basic guitar driven
> rock and roll (rockabilly, CCR styled), some pop. The music is all
> pretty dynamic, compressed and loud, it's certainly isn't chamber music,
> or music with any quiet passages.
> Alan
>
>
>

[xy]

Mr. Dorsey, for resolution purposes, do you support the idea of mixing
on an analog console from the DAW? or do you tend to stay all digital


>
> Every time you change the gain, whether it's normalizing or adjusting it
> manually, you are losing precision. So you should have left them at the
> low level until the final step.
>

[Scott Dorsey]

xy > wrote:
>Mr. Dorsey, for resolution purposes, do you support the idea of mixing
>on an analog console from the DAW? or do you tend to stay all digital

Depends. Staying digital is certainly a better idea from a sonic perspective
since you're avoiding those converters. The problem is that your effects
and processing are limited to digital stuff now.

If you're doing fairly minimalist stuff where you're using no EQ, you're
only summing a few channels, and you aren't doing much gainriding, using
no external effects, staying digital for mix is a huge win. When I do
classical stuff that is multitracked, this is how I work.

Personally, though, I find mixing on typical digital consoles (or even worse
on workstations) is irritating as hell, and it basically eliminates all
the usual processing stuff that you like for production. So when I am doing
anything other than minimalist classical stuff, I use a conventional console
because that is what feels better to me. Then again, most of the time that
I do that, I'm working from an analogue master tape too.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."