This is the single biggest problem we see with mixes submitted for mastering.
There are two factors to consider when exporting your final mix for mastering:
A. Peak level - The absolute highest level the output signal reaches.
B. Dynamic range - The range between the highest level and the lowest level of the output signal.
The loudest part of your song (peak level) should be around -3db to -5db (below 0 level).
In your mix, never let your levels go over 0db on the stereo out (main out) meter, AND never go over 0db on any of your individual instrument tracks or vocal tracks. Remember, loudness maximization is done in mastering, so there's really no benefit to go over-level in mixing. Note - For the record, at times you can peak at 0 level or slightly over if the peaking parts are quick hits like a drum, hi hat, etc. and they don't distort. But again, there's no reason to go over-level.
Sometimes we receive songs that correctly peak under 0db, but the vocals or certain instruments are distorted because the "individual tracks" were mixed over-level. Nothing should ever go over 0db on any channel of your entire mix.
We get so many over-level submissions, I've decided to show graphics of how your .wav or.aiff files should look.
This .wav file peaks at -6db. A little low but just fine. We can work with this.
This .wav file peaks at -3db. This is great.
This .wav file peaks at 0db, but is fine because there are no distorted flat
spots in the wave. The high hats are hitting at 0db and everything else
drops down, leaving good dynamic range. We can work with this.
This .wav is +3db over level. All the flat spots you see are distortion. You can't
just jack up your levels past 0 and distort everything to make your song louder.
Do not submit a file that looks like this.
Once a week, we get at least one file that looks like this! Let's call this the
"trainwreck file." +6db over level and totally distorted from beginning to end.
We can't do anything with this. No one can!
This is an interesting file. We get this quite often too. It's the "train-wreck file"
we mentioned above, gained down -9db. It's completely distorted, but its -3db under 0 level.
A file where the mixer channels are over 0 level, but the main stereo out
is -6db, would look exactly the same.
We can't use this either. Once your song is distorted and over level, you
can't just gain it down. All you're doing is making a distorted file quieter.
You have to be under 0 level on the stereo out, and on all the mixer board
channels when exporting the original .wav file.
Unfortunately, many times the recording studio gives you the "train-wreck" file
and you're stuck with it. Gaining it down yourself does nothing.
As a song plays, the output meter constantly goes up and down. Headroom (dynamic range) is the area between the peak level (when the meter is up) and the low level (when the meter is down). Roughly, 3db to 5db of movement between the high and low meter level is good, but this depends on the genre of music. With a compressed/limited or over-level mix, the meter will not go up and down more than 1db, if any.
The reason we need headroom is because compressed/limited or over-level mixes already have music content removed from them, which greatly limits what we can do in mastering.
It's comparable to a woman who goes to a hair stylist with only 3 inches of hair. The stylist would be “very” limited with what she can do! 12 inches, the sky's the limit.