Hi, guys.
 
So I've got a vocal track with some fairly dynamic sections (some parts too quiet, others recorded too hot), and before I get into the weeds with any manual gain adjustments, I wanted to run Normalize on the whole track just to get the loudest parts down a desirable level (without using compression).
 
I was under the impression that the way normalization works is that it scans all of the selected audio, finds the loudest peak, determines the needed amount of dB by which to meet the desired goal, and then adjusts all of the audio evenly by that determined amount... right?
 
Well, after I ran Normalize on the track, I found that some clips are not being adjusted by the same amount as others...
 
For instance, there are two adjacent (but separate) clips. After normalization, one of the clips went from -13.6 dB to -15 dB (my target peak). That's a reduction of 1.4 dB (or 10.29%). However, the other clip went from -16.9 dB to -19.5 db, a reduction of 2.6 dB (or 15.39%). Clearly not the same amount of reduction.
 
So what's going on? Is the normalization not working correctly (perhaps because the clips are separate)? Am I just not properly understanding how it's supposed to work?
 
Any input is appreciated!
 
 
SONAR X3e Producer
Win 7 64-bit

Nope, normalize doesn't work that way. What it does is raise the level across the board so that the highest peak hits your target level. The relative volumes don't change.
 
SONAR does not contain a tool for doing what you want. There are some third-party plugins that attempt it (e.g. MAutoVolume from Meldaproduction) but they'll only get you into the ballpark.
 
FWIW, this is my two-step method. First, I insert BlueCat's Gain plugin and then automate it so each clip or phrase sounds about right to my ear. Then I apply compression after.

bitflipper
The relative volumes don't change.

 
See, that's why I'm confused, because the relative volumes DID change.
 
Before normalization, the difference between the two clips was 3.3 dB. After normalization, it was 4.5 dB. That's what originally tipped me off that something wasn't right.
 
Your explanation of how it's supposed to work is exactly how I thought (just put much more concisely than my attempt to spell it out).
 
So what could be going on?

Hornet's Autogain Pro will do this and it's cheap.

Thanks, Clint. But I'm not trying to do anything advanced here. Sorry if my original post wasn't clear.
 
All I'm trying to do is adjust the volume in a track so that the highest peak hits a specified point.
 
To the best of my understanding, that is what basic normalization is supposed to do. But for some reason, in my case, it's not adjusting the levels by an equal amount across the board. I'm guessing it's because there are numerous separate clips in the track and I'm trying to apply normalization to all of them at once (if I do them one by one, every clip will reach the specified peak level, and that's not what I want). I can bounce all the clips into one, and then it works right, but I wish to keep the clips separated.
 
I guess I can achieve the same result (non-destructively) by adjusting the track's Gain knob by the necessary amount for the peak to hit the desired level. I was just looking for a way to do it through processing. It probably wouldn't be a problem with normalization if my track wasn't all chopped up.

You would probably have to bounce the track to create one region then apply normalization.

"there are numerous separate clips in the track and I'm trying to apply normalization to all of them at once"
 
That would be my guess. The normalization is affecting each clip individually based on each clip's peak.
You could Bounce To Clips, Normalize, then split the long clip into the original shorter clips. 

57Gregy
The normalization is affecting each clip individually based on each clip's peak.
 

 
You would think that would be the case, if it didn't affect all the selected clips based on one universal peak, but I can't make heads or tails of what it's actually doing. One clip will see a reduction so that its peak hits the desired level, but then the next clip is decreased by a disproportional amount, and comes nowhere near the peak target. *Shrug*
 
 

57Gregy
You could Bounce To Clips, Normalize, then split the long clip into the original shorter clips.
 

 
I considered that route, but it just feels like too much extra work for something that ought to be so simple. I'm thinking I will just adjust the track's gain via the gain knob to conform to the target peak level. At this point, that seems like the most straightforward, efficient, and simple method to achieve essentially the same result.

Another question. Where in the signal chain is the normalization applied? If different clips are routed through effects or busses/patch points that may involve some gain adjustments is the normalization applied pre or post? If any of the intervening effects are using a gain adjustment with a different algorithm than the overall normalization that could cause unexpected results. It sounds like the safest thing to try is to normalize only a fully rendered project, but then, it would be an enormous amount of work (or impossible) to try to get it all unscrambled again.  

I'm almost entirely certain normalizing is done on the raw wave data pre everything. It would be half useless otherwise.