[MPA] "Mix CD" Consistent Loudness

[Scott Jacob Loehr]

>Hi,
>I'm wondering what people are doing in the following situation. Say 
>you are making a "mix CD" (or even a CD of your own music) where the 
>tracks are all taken from different sources, and hence have 
>different apparent "loudness" or volume. What approach do you use to 
>make all the tracks have the same loudness? Is there an automated 
>method, or do you simply listen to all the tracks over and over 
>again and tweak them until they are what you consider to be "equal 
>loudness"?
>-jd

	I do all my work in MOTU's Digital Performer (24-bit, usually 
44.1 k), and I've frequently had to put tracks from different sources 
(cassette, DAT, CD, reel, etc.) and studios onto one CD.  I usually 
record each stereo file onto two mono tracks (panned hard L & R) to 
match levels (lots of stuff isn't mixed or copied correctly).  I 
often have to draw in a lot of volume data via automation to get each 
stereo pair to sit more or less correctly, and I'll use a master 
fader to boost & cut overall volumes for each tune.  I'll apply a 
seperate EQ to each mono track or stereo pair if needed; rarely does 
one EQ setting work for all tracks when they're from many sources.  I 
never use compression; I don't like what it does, especially to final 
mixes.  If I'm just going after volume adjustments, I'd rather draw 
it in manually.  Once I'm happy with overall volumes, EQ, and the 
spacing between tunes, I'll play back the entire file and check for 
any peak overloads.  If there is more than 6 dB of headroom at the 
loudest parts, I'll raise the levels on the master fader by a few dB. 
Then I bounce to stereo (still at 24-bit), and I open the file in 
either Spark or SoundDesigner2.  Here's where things can get really 
tedious:  I'll search for the peak of the whole stereo file and 
evaluate whether or not I can re-draw the waveform to decrease the 
volume.  If it's an isolated transient, like a drum hit or some kind 
of plosive sound effect, I can usually smooth off anywhere from three 
to ten dB.  If it's any type of sustained sound, I don't go any 
further, other than to perhaps gradually lower the volume during that 
section, though this usually only nets me two to four dB of headroom. 
I then bounce the file down to 16-bit.  I open up the file in Jam, 
and I test-normalize the bounce, then reduce the gain by 0.1 dB just 
in case.  I save the Jam file as an Image File and press a test CD, 
then I listen to the final in three environments:  headphones, my 
studio monitors, and in my car while driving (generally on the 
freeway with the A/C running).  I realize that this entire process is 
very time consuming, but I don't like cutting corners.  If I need 
further assistance, I'll consult with some of my other engineer 
friends in town.  Worst case scenario, I could always send the file 
to a professional mastering facility, if the budget allows.
	Of course, if I'm just making a "mix CD" (or "LCD", as I like 
to call them, as in Lowest Common Denominator with regards to 
quality, such as mp3 files), I'll just normalize each song in Jam and 
keep my fingers ready to adjust the playback volume knob while I'm 
listening.

Scott Jacob Loehr