Track audio normalization

I'm recording tenor sax into a track, and normalized the audio to -0.1 db.
 
My track level meters show peaks of 2db throughout in playback.
 
I normalized to -2 db, and get 1db peaks (overload) in the track.
 
How is normalization working?  Or is it NOT working properly.  I'm assuming this is not an average vs. peak thing.
 
How is one to control anything if normalization can't be counted on to do just that?
 
Hope I'm overlooking or ignorant to something.

Hmmm.

Brundle,
 
What a great response.  I'm all ears and gratitude. 
I checked back on the project, and indeed the person who recorded me had added EQ boosts on the ProChannel EQ to add some rasp to my horn. 
 
WHEW! That mitigates a HUGE worry I had that there was no reference for 0db in Sonar's normalizing.  That would seem to be like sailing the ocean without charts, mix-wise.  I was SO afraid that it was a non-sensical program blunder, because it would mean that fundamental issues were up in the air.
 
Now, to your final question about "Why normalize?":  Maybe ignorance.  In this instance I was recording a practice tenor part just to feel out the other tracks done by other artists.  The level was averaging about -20 db due to low settings on the mic preamp on the first takes, so the person just normalized my track to get the levels up.  If we were doing final tracking, we'd get the front end right, level-wise, at the preamp, typically peaking at about -6 db.
 
That being said, how would you do it if you were stuck with a final that was too low?  Just boost the channel volume?  I'm all ears, and thank you for your response.  It made so much sense that I went right in and found the EQ boosting with great relief.
 
Jon

You just epitomized the spirit and purpose of a technical forum, Brundle.  I thank you.
 
I take it that using the Gain atop a track avoids little complications like post send levels and whatnot that would be in play (pun intended) when using the channel strip volume control?
 
Best wishes to you!
 
Jon

normalize is a tricky function.  Sound Forge has several presets which tell a lot.  Normalize for music is -18 dB, I think, while VO is -6 dB.  As you found, a -.1 is way to high and is guaranteed to sound bad, esp. bringing it up from very low.  Overs shouldn't happen, but it will still likely sound like crap and you'll have to lower it back down.
 
The gain is a much better option, tho I've been known to normalize truculent tracks.  But only a few dBs.
 
@

There's nothing inherently wrong with normalizing files, just don't normalize to -0.1 dB. It doesn't leave you any room for further manipulation, such as an EQ boost. Even normalizing to -3 dB puts you in pretty much the same boat. -12 dB is probably a more practical ceiling. AT's suggestion of -18 dB is a good target.
 
One of the issues, as you discovered, is that the normalization process looks at raw sample values, not "true" peak values, which are often higher. That's why when you normalized to -2 dB you subsequently saw 1 dB peaks. That can be avoided by keeping sample peaks below -3 dB, but that still leaves too little headroom for mixing.
 
All of this is really moot, though, since there are only a few situations where normalization is warranted. If, for example, if you had multiple takes of the same instrument at wildly different volume levels, normalization could quickly get them into the same ballpark. Even then, you'd probably only bring them up to peaks between -18 and -12 dB.

We digitize thousands of tapes and vinyls yearly with Sonar. We have already normalized 100.000 times.
This all works perfectly with Sonar since over ten years.
But it must in no case any plugin in the effect-bin.  And the gain and volume must be set to "0".
Otherwise plugins and gain and volume change the level.
When normalized to -2db all other audio-software will show -2dB. That works absolutely perfect without any fail.
 
 

What I found somewhat startling to realize is that making a cut with an EQ can also result in clipping/increased peaks. In effect, if you cut a frequency but that frequency was actually inhibiting another peak, you're creating a higher peak. It's not usually a big issue but if you're skirting so close to 0dbfs it can come into play.