To me, mixing at a constant calibrated level is very important. I rarely change the volume of my speakers during a mix from start to finish. So why is this important? Keep reading...
The reason behind this is how our ears work. Basically, we hear frequencies differently at various levels. I'm not going into the science behind this but if you want to read more, research the Fletcher-Munson Curves.
In a nutshell:
At low listening levels we perceive the midrange to be louder while the low and high frequencies is less pronounced. At high listening levels the low and high frequencies will seem louder while the midrange is less pronounced.
As you can see, monitor levels plays a big part of how we perceive frequencies and keeping your monitor level constant, at a level where our ears have a flatter frequency response will help you mix faster and with more consistent results. You will also soon learn when the tonal balance of your mix is right.
According to the Fletcher-Munson curves our ears have the flattest response around 83-85db, however for most studios I find this level way too loud. Sound of sound magazine published a table in this article with recommendations based on room size. My room is about 110cubic meters so according to this they recommend a reference level of 76db which I feel is more comfortable for long periods of time. In smaller rooms you might want to go even softer.
The calibration procedure:
1. First you have to decide at what digital operating level you want to mix at. I mainly mix pop and other contemporary genres that usually end up with fairly little dynamic range, but I still want my mix to not be too squashed (before mastering) and still have some dynamics and punch.
I use LUFS (Loudness units relative to full scale) and set my operating level to be -14LUFS. I feel that a reference level of -14LUFS is optimum for what i'm mainly mixing, and that gives me enough headroom for louder short term peaks, the dynamics is controlled but not totally squashed.
2. Next you need a test oscillator with pink noise settings and a LUFS meter, most DAW's have one. Make sure the pink noise is set to RMS and that it is playing in mono.
Then play the pink noise, look at the LUFS meter and adjust the pink noise to the level you decided in step one, (in my case -14LUFS)
NOTE: RMS and LUFS is not the same thing and for the signal generator to output exactly -14LUFS it has to be set to -14.3RMS. This is because LUFS (with ITU-R BS.1770 algorithm) takes into account that our ears is more sensitive to certain frequencies, among other things I'm not going to get into. This is not really important as long as the pink noise is outputting -14 on the LUFS meter.
3. Now you need an external SPL meter set to C-Weighting and Slow mode (averaging mode). These can be bought for about £20-30, or there's iPhone apps that can be accurate enough.
Then play the -14LUFS pink noise, place the SPL meter in the listening position and turn up the volume of your speakers until it reads 76db (or the level you choose)
That's it! Now you have monitors that are calibrated to the loudness of what works best for your room your and the music you're creating. You'll find that after some time you'll be able to judge when a track is too loud (squashed), too soft (too dynamic) and when the frequency response is just right.
When mixing I try and get the mix to hover around -14LUFS on the loudest section of the song.
It's still a good idea to test how you mix is sounding at very low levels and also at a loud level. At low levels I find it easier to hear if something pokes and at loud levels, you want to make sure your mix still sounds balanced, because to be honest, most people listen to music too loud so it's definitely good to make sure it don't sound to bright and bassy at loud level.
To be able to go back to you reference level quickly make sure you mark you monitor controller where your reference level is. I use the Kii Three monitors and have to luxury of being able to set zero on the Kii control to be my reference level.
NOTE: this calibration procedure only applies to mixing. When mastering I just turn my speakers down by the same amount of limiter gain I'm applying. Which will roughly be 76db.