AW: Mastering music

["Rainer Thelonius Balthasar Straschill" <rs@moinlabs.de>]

> I guess you're talking frequency bands here when saying "in 
> the middle of the dynamic range"? 

No, I'm not. I'm talking "dynamic range" here, meaning:

A normal compressor does the following. It completely "ignores" signals
below a certain signal level (the threshold), signals passing above it will
be subsequently attenuated by a certain factor (the ratio) - and this
happens up all the way to 0dB.

Another way to go is to have a range instead (say -30dB to -20dB), and if
the signal is in that rage, it gets attenuated according to your factor 
(say
e.g. 2:1). So a signal with -30dB goes through as -30dB, a signal with 
-20dB
goes through as -25dB. So now comes the tricky part:

If the signal goes beyond -20dB, it doesn't get compressed any further - so
in this example a -5dB signal would come out at -10dB etc.

This way, you compress dynamics in the midrange of the dynamic spectrum 
(the
soft parts of your song, which often get lost under suboptimal listening
conditions), while the dynamic transients in the loud parts (read: drums)
get preserved.

Multi-band compression is another thing to which a lot of professional
mastering engineers react quite adversely btw. One reason is if you do 
this:

> separate sub channels. Each sub channel is then armored with 
> a low cut and high cut filters to enable only a certain 

these eqs have quite a frequency-dependant effect on the signal's phase, so
you get a phase washup all the way. For that reason, Waves did the LinMB
compressor which compensates for the eqs phase shifts.

        Rainer