Audio mixing is the most important part of any production. It is where all the elements come together, and when we get it right, it pays off dividends like no other discipline can in filmmaking. I’m including a segment from my book on Live Audio Mixing herein, to help explain how “compression” can be one of the most misunderstood and important tools in the mixer’s tool belt.
Even though the book is for live music mixing, the concepts and application of compression for post production
mixing is identical. This begins on page 163 if you want to follow along in your copy of the book. Enjoy.
You want your mix to sound hot? Sizzly? Killer? Me too.
First, you have to start with killer, sizzly and hot musicians. Don’t have them? Sorry. Well, you can at least get them to sound smoldering, bubbly and dangerous. How? Simple.
Of course, Equalization and all the other tools in this handbook are important, but in my book (and this is my handbook), compression brings the pop, the smack, and the in-your-face sound that we love both live and in the studio. Want your mix to sound clean, studio quality and free of noisiness? Me too. It’s simple to get this too! Expansion. The good news is these processes (and their accompanying processors) are easy to use and reasonably inexpensive. The bad news is that using the two processes together tends to cancel each other out. This said, in the crucible of your mix, like a piston firing in an internal combustion engine, without a spark, you’ve got no power. These two tools are the spark that will ignite your mix and put a silly grin on your face.
What is compression? Well, quite simply, it’s a process whereby a signal is not allowed to rise above a certain level. It seems a pretty sterile and useless definition where creating a powerful mix is concerned, I know. For years I thought compressors were stupid and useless. Without getting into the nitty-gritty of why they work, I’m only going to address recipes for getting certain sounds. It helps to look at compression as an automated hand on a fader…one that you don’t have an extra appendage to utilize as a homo sapien. This hand literally moves this phantom fader up and down in an effort to keep the volume of a signal from ever rising above a certain point. That fader isn’t located on your console, it’s a magical fader located in the box (or section) of the compressor. You determine the level beyond which the sound cannot go, what tolerance you will allow relative to that level, how far it can move the imaginary fader, how fast it moves the fader and possibly even create a new sound level altogether. Handy device.
The basic front panel of every compressor includes the following controls, and although some controls are collapsed into one, they should be there on any compressor that is worth its salt.
Usually they are indicated as 1) Threshold (the level above which a signal begins to be attenuated), 2) Ratio (the force that is used to resist the signal once it crosses the threshold), 3) Attack/Release (the speed at which the signal is attenuated (above the threshold) and released from attenuation once it goes below the threshold), 4) Input (how loud the incoming signal is) and 5) Output (how loud the attenuated signal is). Here is an example: You have a flubby bass drum sound.
What if you could know within microseconds what kind of level the drummer’s drum smack was going to be? What if you pulled the fader down 5 dB 3 msec after he hit that drum? You would end up with a loud attack that dovetails quickly into a nice sustain (for a bass drum that is). If you were to turn that signal up (since the overall sound of the hit would be lowered by 5 dB), you’d have a smacking great punchy bass drum, right? What about an electric bass that is lethargic and the player is too dynamic and keeps sonically jumping and ruining your mix? In that case, if you knew when he was going to get loud, you could jam the fader down 3-10 msec after he played a loud note. This would give his loud notes a good smack, and yet the overall volume would be at the same volume as the soft notes. Cool, eh? Well, friends, this is what compression can do. Automatically. Easily. It also adds a certain power to the frequency of certain sounds that our ears have grown used to hearing as “edgy.” You, obviously want to be careful with too much use of compression, because too much of anything is no good. More than EQ or effects, the quickest way to get a sound to pop, or at the very least to get it under control is to drop a compressor into the signal path.
In the examples I’m going to give you, it is important to think of the threshold control as the “effect mix” pot. In other words, if you want more compression, turn that pot down (into lower numbers). If you want less compression turn it up (towards 0 db). Remember, this control sets the level that the input signal should not go over. Thus, if you set a threshold of digital –5 db, almost no signal is going to go up there to get compressed because -5 db is pretty loud. If you set it at –50, in 16 bits (digital), nearly all of the listenable signal is over that, thus you’re going to have everything compressed. Make sense? No? Just try it and you’ll see that it works this way.
Take the example in the diagram. The sound wave displayed is that of a sampled snare drum that has little snap to the attack. It’s your basic “blocky” sounding pre-effected sound. At the top is the nature of the snare sample without any compression. Even at first glance it’s apparent that this sound is extremely transient and has good sustain…for a snare drum…which means nearly none. All of the following examples have an extremely draconian threshold of -23 dB and 30:1 ratio of compression. Normally it’s not necessary to use this much compression, but for the purposes of visually demonstrating an audio example it is required. Also, it is important to note that all of these examples have output “compensation” on: the output is kept at a relative level so that the draconian compression we are using can be better seen visually. Otherwise, it would be very difficult to compare examples. With the second example, the 0 ms attack really chews into the already crushed attack and takes away whatever is remaining of its crack. Now it’s almost a bump-like attack. It also lengthens the sustain of the snare significantly and actually provides a beautiful “body” to the sound…but there is no attack of any kind. Now this can be good if the attack is too much, but terrible for our needs here. The 11ms attack and 90 ms release example demonstrate that the longer release actually lessens the sustain of the snare while exacerbating the crack in the extreme. For our purposes, this is excellent, but we’d like a bit more body in the sound. The final setting of 41 ms attack and 5 ms release shows a good balance between the previous two examples whereas the attack is accented while the sustain of the snare is spared. It is important to note that the compressor used in this example is a digital compressor and has the ability to return the overall volume for what would normally be a signal rendered quite a bit softer from extreme compression. This is why the über compressed second example looks louder than the rest. In actuality an analog compressor would have rendered it the softest. Still, the effect of compression is not lost and the overall volume (even with return of compressed signals) remains constant and under or at our threshold.
If something sounds flubby, lethargic, or obese when it should sound snappy, you want a setting that looks like the following: threshold set to a level that is around 2-3 dB above the overall loudness of a sound (as opposed to its initial attack), set the ratio to 8:1, attack to 10 ms, release to 50 ms and adjust your output accordingly. In the case of a drum hit, set the threshold around 10-15 dB below the level of the initial attack and the release to 10 ms. If it is still too fat, then raise the ratio and/or lower the threshold. Be careful not to lower the threshold into the volume of the sustained portion of the sound. With the setting this harsh, it will create unpredictable results…not to mention feedback hell.
If something is far too dynamic for your taste (like a piano, electric bass, trumpet or vocalist), do the following: set
the threshold well into the overall level of the sustained portion of the sound, ratio 3:1, attack 50 msec release 750 msec This should control the level of the signal without coloring the sound too much. If it doesn’t seem to be doing enough, lower the threshold and/or raise the ratio. If, in the case of a piano, the attacks of the notes are too hard, reduce the attack to 0 msec. If you’re looking for that punchy vocal sound, raise the ratio to around 25:1 lower the attack and release to 0 msec. You’ll get a sound (depending on your threshold) that will knock the backside off a concrete elephant. You’ll also get increased sibilance and you’ll lose some top end off the spoken or sung part of the sound (just trust me on that seemingly puzzling or enigmatic sentence). Use your EQ accordingly. You’ll also have more feedback trouble than you’ve ever had to deal with before. Be careful.
Now a lot of people get a little shifty-in-their-seats when I start talking about the liberal use of compression in a live mix. There are some drawbacks; namely the alarming rise in the probability of feedback. Again, without getting into the nitty-gritty of why, using compression brings up the volume of a signal without it sounding louder…but a microphone doesn’t care! As a rule of thumb, for every decibel that you compress (and recover the volume loss by raising the output) you’re adding a decibel to the signal’s level as far as feedback is concerned. So, if you know you get feedback at +5 dB and you’ve been riding your fader at 0 dB, when you compress the signal 6 dB and volume compensate: Voila! You’re eyebrows get ripped off. It doesn’t always work this way, but feedback tends to show up when compression becomes more draconian.