Klas Strandberg wrote:
> I agree! Except sometimes you have to. When I make an MP3 file to put on =
> site, I know that lots of listeners can't adjust their replay level.
> And the the MP3 conversion I use seem to work best when peak level is -6d=
> don't know why.
> Otherwise I think you are right.
> But I would still like to get a definite answer: How much damage is there=
> for example when normalizing? It feels wrong, not even knowing. =
In some ways it's a tempest in a teapot, and other ways it's not. The
listeners also have the equivalent of our normalizing, called a volume
control. As I understand it each listener has a preferred decibel level
to listen to recordings. They will adjust the volume control to obtain
What matters in producing something like a CD is that the listener does
not have to change that setting for each track. That does not mean every
track should measure out to the same level, the tracks should, in some
way represent the variations in sound level that are in the natural
environment. Normalization of the different tracks to the same level is
probably undesirable, but so is having the tracks not work together.
Having to ride the volume control is one of the things that will quickly
turn a listener off.
As a farther ideal you can extend the idea of a CD's tracks working
together naturally to all the CD's you make working together naturally.
As for me I don't normalize final output to a setpoint, I listen to what
I get and balance that. I do generally do sort of normalization before
processing the sound. Both in choosing the settings when I record, but
also by adjusting the level. I find I can get more consistent results
out of filters if they are working on recordings that are at a similar
level. It's not just in mp3 conversion you can find there is a optimum
Since with most filtering we are removing sound energy, there is a
tendency for the signal to drop in each step. I compensate for that too