Updated: Oct 31, 2019
You may have read the post about how DAWs destroy your recordings (if you haven't, there's worse ways of spending five minutes), so you already know that - when using a digital recording system - the right thing to do is to set gains so that levels hover around the middle the dynamic range scale (say an average level of -18 dBFS and peaks around -12 dBFS).
And if you've read my post, you'll have set up your DAW GUI so that its visuals match the good quality of your recording.
It's a fact of life, however, that most music we hear is mastered to reach the top band of the dynamic range, i.e. peak at or a little lower than 0 dBFS.
CD masters, which are still often peak-normalized, are meant to be printed to CD, and to stay there (no ripping!).
Therefore they tend to be mastered so to peak at 0 dBFS, because the digital/analog converters of good CD players and a good downstream playback chain will handle that level just fine.
Even loudness-normalized masters may well still have peaks well up on the scale: masters meant for streaming will often peak a little lower, from -1 to -0.5 dBFS, to counter the effects of the inevitable aggressive compression that streaming services apply.
Still very high levels.
Ok - I hear you say - so what? Why's all of this is relevant?
Well, because this means that - especially when referencing but also when recording - you will often have the big knob of your monitor controller set up so that these commercial tracks play at reasonable level. Or the interface volume knob; or the Windows Taskbar volume slider... you get the gist.
Not too loud, not too low. The volume is set for listening to "Beat it", so that you get the snare, the details and all of Van Halen's small pick scratches just right.
True, If you put on something from the 80s you'll probably turn the volume up a fair bit, and Death Magnetic will get turned down a lot, but basically your volume knob will be set for optimal listening a track mastered at full-scale.
But with the volume set up that way, what will happen to your working mix, or the monitor mix in your headphones when recording? If you're working with 24 bits digital recordings, and doing it right, neither are going anywhere near -10dBFS, let alone 0 dBFS, so...
It will sound puny.
As in really-low-volume-puny.
The risk here is that the inexperienced engineer, upon hearing such low volume, will conclude that his or her tracks have too low gain and will raise the preamp or channel gain to get a more satisfactory level in monitors or cans.
Pity the fool.
The right response here is leave the gain as it is - which is just where it should - and raise the playback volume instead!
If you are uncertain on what is the difference between volume and gain, have a read of this post..
The listening volume is the result of both the amount of channel or preamp gain (which we need to set so that signals hover around middle scale) and the playback volume knob.
So if you're recording at averaging at -18 dBFS and your signal seems low, raise the playback volume.
Before blasting "Beat it" again, lower it down.
For referencing, either you change the volume manually (boring!) or you can use one of the many A/B plugins that allow you to maintain the same volume between your unmastered mix and the mastered track. They're very simple and extremely useful.
So here's your conclusion: when recording or mixing, set the gain so that your levels hit middle scale.
If you feel your volume is too low, stay away from the gain knob.
Reach the playback volume control, and pump up the volume instead!