Just back from a new mix and ready for what will be my last installment on the art of mixing. To restate again, these are just some of my favorite techniques that I either use often, or solutions to rare problems that crop up. By no means is this meant to be exhaustive.
One of the hardest things for beginners to gauge, is the use of compression, so it seems. I have heard so many amateur tracks that are "overcompressed" it isn't funny, lending strange artifacts to your tune. Again it's a case of training the ear to gradually recognize small changes, and to identify what it is you are listening for...hence the tendency for the beginner to turn up the compression until they can "hear it"...which is usually WAY too much.
How much compression? I find that when I've been away from recording a long while, it's hard to recognize a change of less than 4db of level...very soon however, I get my "ears" back and 1db starts to sound like a lot...This is important, because when you gain this skill, you can listen to level fluctuations in a track, and say to yourself, "That sounds like it's about a 2db difference between the loudest and softest parts of the track" and so you can begin by setting your compressor to about -2db and see how it goes.
Now ratio is another thing, and here's where I use my 1st technique...If I have a track that needs a lot of compression (one that fluctuates level-wise, or needs good sustain) like bass, vocal, etc. I NEVER compress all at once....I will compress at a low ratio in the first stage (say, while recording) and then re-evaluate and compress again harder in succeeding stages (bounce and/or mix) possibly all the way up to 100:1 ratio.
Percussion is a different matter, and in some cases will record with high ratios right off the bat, however, depending on what kind of sound I'm after.
Another good thing to know is ways to use EQ to create an "open" mix, that is, one where instruments occupying the same frequency ranges aren't running over each other.
I'll give an example using 2 tracks: a stereo string section, and a single vocal (panned center). Now, the timbre of the strings roughly occupies the same frequencies of the human voice (say, from tenor to soprano) so they will tend to fight each other in the mix...you are faced with a dilemna; change the relationship (level difference) between them raising the voice (or lowering the strings) or vice versa.
Detail of the lower track will be lost in the process.
Here's your alternative: Each part has it's own "active" frequencies. The voice is a complex signal (same with strings) that covers a lot of frequencies. The voice will have upper frequency that gives it definition and makes the words clear & audible for instance, and lower region where the "warmth" lies, etc. There may be others as well: breathiness, resonance, etc. (These differ with voice range and whether the singer be male or female).
In the same way, the strings will have "bow sound" (for lack of a better term), and their own resonant ranges. If you are lucky, these ranges will not coincide, so that a small (never more than 2db) boost to one tracks "active" frequencies and a corresponding (if needed) cut to the other track will produce a nice blend without losing anything from either track. Or boost both tracks active freq...Or boost/cut symmetrically both tracks...the options are many, the theory the same.
Lastly, I will share a technique I seldom use but have found of value today. I always try to make my mixes work so that the things will blend nicely without a lot of fussy level changes when mixed...However, this is not always possible.
Today, I made a very complex mix that was very full with many parts, virtually every frequency range spoken for. Try as I might, I couldn't keep the nice "fullness" of the vocal, even using EQ, compression and other FX.... Lowering the midrange instruments (everything above the drums & bass) made the vocal clearer, but at the expense of the rest of the piece.
So as a Last Resort, I used a technique known as "ducking"...When the instruments were playing I allowed them to run at normal level, but whenever the vocal entered I would ease the fader(s) down 1db until the vocal ceased...and so on. Not noticeable to an untrained ear, but a world of difference as far as the clarity of the vocal and the fullness of the mix.
Well, that's enough of "fish" stories. Engineers can sit around and talk about the problems they encounter and the solutions they come up with virtually forever. But this blog is general enough that to continue I would to risk becoming a bore! So let's leave it there...
