dgatwood said:
With the caveat that there's also something called RMS normalization
...
Note: I don't approve of that misuse of the term normalization.
It's amazing how long it always takes for someone to mention RMS normalization. That's not anymore a "misuse" of termonology as it is an accurate and appropriate description of what the app does. The misuse in terminology, IMHO, is the use of the term "normalizer" without specifying which type it is. There are "peak normalizers" and "RMS normalizers". To use the term "normalizer" without a modifyier is almost as bad as specifying dB without specifying which dB scale you're referring to. (That last one drives me insane, BTW
)
the question is why are you using the gain or normalize plugin? are you not recording hot enough (but not too hot). 60% of 0dB should be fine in a 24 bit environment. Remember, the average level of a track should be around -18dBFS. Gain or normalize is a last resort for me. For example if I have something recorded from someone else that comes in at around -40dB or something...and I just HAVE to bring it up louder.
More to that point, peak normalization was never originally intended to be used in the manner in which we describe it, particularly in a multitrack environment. Both kinds of normalizers (peak and RMS) are great for use on multi-take or multi-clip single mono tracks - musical or non-musical - which is what they really were originally designed for. (We tend to forget that audio NLEs have many applications beyond the music recordings we are mostly into on this board.) And subsequently, we musical-minded people have learned to adapt them to use sometimes on our stereo mixdowns as part of pre-mastering.
But in a multitrack environment, the main use of normalization - both peak and RMS - is to try and
match levels of multiple clips within a track, not to rase the volume of a single track.
This is why they let you choose levels other than 0dBFS. There's *nothing* anywhere that says that normalization is intended to raise the volume of a track. We all just assume that because of an idiotic "turn it to 11" mentality. In reality, normalization (think of that name) is meant to be a process for *matching* levels between multiple clips. That could mean bringing the volume of clip A
down to the volume of clip B just as much as it means bringing the volume up. You are "normalizing" the volume of the individual clips - i.e. making them all sound pretty much the same. That's what they are really supposed to be for.
What makes them superior to gain controls in that respect is that one can choose a specific overall level (either peak or RMS) and then just run each clip to that level. With a gain control, one would have to raise or lower each clip individually by it's own amount to provide matching levels. Yes, peak normalization and gain control perform basically the same operation, but for the intended purpose of matching clip levels, normalization is much more automatic.
The big irony is that for that purpose, RMS normalization is actually usually more effective than peak normalization in getting the average job done, especially when there's signifigant differences in the sonic density of the clips to be matched.
G.
