Unlike an expander, which increases dynamic range, a compressor reduces dynamic range. In recording, running a signal through both a compressor and an expander can be very effective.
Why would you want to reduce and enlarge the dynamic range at the same time? Actually, you wouldn’t.
They don’t both come into play at the same time; an expander does its thing when signals are at their quietest (or nonexistent), and a compressor does its thing in the louder part of the dynamic range.
So if you’re recording a cymbal crash through both a compressor and an expander, the expander will be working before the sound begins; the expander’s gate opens up immediately when the cymbal is struck, and then the compressor takes over.
The compressor works perhaps for a few seconds while the cymbal decays (with the expander doing nothing, since the signal is over the expander’s threshold).
Then the signal enters a kind of no-man’s-land, between the compressor’s and the expander’s active ranges, where neither circuit does anything to the signal.
Finally, when the expander senses that the crash is decaying below its threshold, its gate begins to close again (see Figure 1).
This process accomplishes two things: the expander cleans up the noise before and after the crash, and the compressor tames the initial peak and thereby allows the whole signal to be brought up in volume, allowing it to have more punch and presence in the mix.
If you were recording or sampling a series of cymbal crashes, one after another, the compressor would be even more beneficial: It would tend to even out the crashes in volume, which would make the quieter crashes less likely to get buried in the mix and the louder ones less likely to overwhelm the mix.
As a bonus, compression makes a sound less likely to overload stages downstream in the signal chain—which is particularly important if you’re recording digitally.
Figure 1: A crash cymbal (left), and the same crash cymbal through a compressor/expander (right). The compressor and the expander come into play at different points of the cymbal’s decay (click to enlarge)
Here’s a look at a compressor’s typical parameters and how to use them:
Threshold: To understand how compression works, it helps to imagine expansion upside-down. When a signal rises past the compressor’s threshold, the compression circuit begins to kick in, and when a signal falls below this threshold, the compressor stops working.
So compression happens only when the signal is above the threshold—just as expansion happens only when the signal is below the expander’s threshold.
Given a gradually rising signal, compression can kick in suddenly, which is called hard-knee compression, or the circuit can come into play gradually as the signal rises, which is called soft-knee compression.
Some compressors allow you to specify which kind it performs; soft-knee compression tends to sound more transparent and natural.
Ratio: This term is a little easier to understand regarding compression. In an ordinary signal-chain stage, such as a mixing board’s channel fader, the gain is linear: Any increase in level at the circuit’s input will be matched by an identical level increase at the output.
If it’s a unity-gain stage (meaning that no amplification is occurring), three more decibels going into the circuit will result in 3dB coming out. This is a 1:1 ratio: what you put in is the same as what the circuit pumps out.
A compressor changes this ratio, but only in the region of the dynamic range that’s above the compression threshold.
If the compressor is set for a 2:1 ratio, that means that when the signal level is above the threshold, increasing the level going into the circuit by 2 dB will result in only 1 dB more amplitude at the output.
Likewise, pumping in an extra 10 dB will result in only 5 dB of output. But if the signal is below the compression threshold, pumping in an extra 10 dB will result in a 10 dB increase at the output—the compressor is unity-gain (1:1 ratio) below the threshold.
Figure 2: Above the compression threshold, at a 2:1 compression ratio (left), a 4 dB level increase at the compressor’s input results in only a 2 dB level increase at the output. With a high compression ratio of 10:1 (right), you need to put 10 dB more signal into the input to get 1 dB more output signal (click to enlarge)
Figure 2 shows how this works in graph form. It should be easy to see that if you set the compression ratio higher, you need to pump even more signal into the circuit to get the same rise in output: with a 10:1 ratio, a whopping 20 dB of extra signal level will cause the compressor’s output to rise by only 2 dB.
With an infinite compression ratio, you can’t get the output to rise over the threshold no matter how much signal you put into the circuit. Any compression stronger than about 20:1 is considered limiting.
A limiter is like the flipside of a noise gate—it’s kind of black-or-white, either doing its thing or doing nothing (depending on the signal level at the moment), without much gray area in between.
Figure 3: With both a compressor and an expander in line, the gain is unaffected only between the compressor’s and expander’s thresholds (click to enlarge)
Combining things, Figure 3 is a graph showing how having both an expander and a soft-knee compressor in your chain affects a signal’s dynamics.
Attack & Decay: These parameters are essentially the same as in an expander.
Attack specifies how fast the compressor gets to work when presented with a signal that’s above its threshold, and decay specifi es how fast it returns to a 1:1 ratio when the signal falls below the threshold.
With a soft-knee compressor and a signal that slowly rises and falls in level, attack and decay may not come into play at all—but when presented with things like sudden transients, they can have a definite effect on a sound.