Demystifying Compression

Have you ever wondered exactly what compression does, when to use it, and just what makes it such an integral, yet controversial part of the production process? Some producers get years into their careers without so much as looking at a compressor; this is nothing to be ashamed of, as the effect is very tricky to pin down- it is difficult to conceptualize, comprises many parameters, and spans a wide variance of applications, to say nothing of the results of its use, which can be extraordinarily subtle or blatently obvious. Read on, as Lucius tackles the second volume of this series, with a primer on compression.

In the simplest of terms, compression amounts to a reduction of detail in audio, most often to make something stand out more in a mix. Getting a little more technical, it reduces dynamic range (the span in volume between a song’s quietest elements and its loudest). Though it is not always the case, in most applications, compression is accompanied by an increase in volume, owing to the fact that it increases headroom (the gap in volume from a song’s loudest elements to the limits of the equipment being used to produce it).

Still lost? Try thinking of a single track within a project (synth, percussion, anything) as a dog, and the compressor as a leash. Every dog is unique, just as every production is unique; some dogs do not even need leashes, while other dogs need to be chained to a stake. Know that there exist special compressor types where this metaphor completely breaks down, but for the most part, it will serve just fine.

Makeup Gain
What does it do?
Before going into detail about all the other parameters of a compressor, it is important to remember that a compressor is (in most designs) technically reducing volume. However, many compressors feature a “Makeup” button to bring volume up to unity gain (the maximum volume before the signal begins to digitally clip) or close to it. Sometimes this feature might be labelled “Auto”. This is perhaps the application for which it is most often (or infamously) used in modern electronic music.

How might that sound?
It will make a sound louder. Turn it off, and the sound should get quiet.

And this relates to the dog… how?
Makeup gain is basically going for a walk with the dog leashed. The owner is restraining their dog, but only so the two of them can go places. Similarly, a compressor is frequently used to make the sound more uniform in volume, so that it can be made louder.

Threshhold
What does it do?
Threshold is the volume level where the compressor begins to act. Above the threshhold level, which is usually measured in decibels, the compressor will reduce volume by a magnitude dictated by ratio. The higher the volume would have reached beyond threshhold, the more dramatic the effect will be.

How might that sound?
This is probably the most difficult to pin down, since it is most dependent on the remaining parameters, but it will basically squash everything above its set level. Try running it particularly low and gradually increase the threshhold to get a sense of what elements the compressor reduces or removes, but again, the remaining settings will impact the threshhold’s changes dramatically.

The dog. Tell me about the dog.
This correlates roughly to the length of the leash, though that comparison is not perfect- most leashes lack much of a give when the dog reaches the end, where compressors are much more gradual, but read on and this should make more sense.

Ratio
What does it do?
Ratio dictates the reduction in volume beyond the threshhold level, literally, as a ratio to a single decibel. A 2:1 ratio means that for every decibel above threshhold in the unaffected sound, the compressed sound will be at half volume. A 3:1 ratio means that a gnarly jungle break which started at 9 decibels above threshhold before compression will be brought down to about 3 decibels above threshhold. Most compressors top out at an infinite compression level, at which point they are effectively brickwall limiters, as the sound will get no louder than threshhold. Many limiting/clipping effects are simply compressors with automated attack and release parameters.

How might that sound?
Depending upon the settings of the other parameters, lower ratios amount to a sort of pumping effect, while higher ratios eventually amount to distortion (distortion and overdrive are effectively compression with little to no attack/release time, and potentially more complex reaction beyond threshhold).

The dog. Tell me about the dog.
This is the elasticity of the leash, which for lower settings one should imagine an unrealistically stretchy leash, and for higher settings the leash is more rigid. As the dog runs further away, s/he has to work harder against the leash, just as the signal will get compressed more dramatically as it gets louder. Infinite ratios are like chain leashes for big, rowdy dogs, just as they amount to distorted effects on a signal.

Knee
What does it do?
Knee introduces a subtle amount of compression just before and after threshhold, making the effect more gradual. This is most often tracked in decibels. Soft-clipping distortion is basically distortion with a knee. One should be cautious not to abuse this effect, as it makes an impact below threshhold, and may mess with a signal’s dynamics when it is intended to save them.

How might that sound?
The knee is perhaps the second-trickiest parameter to pin down aesthetically, as it is a function of the threshhold, but it will bring a little more of the dynamic range into compression, and make the initial jumps beyond the threshhold more subtle. This could be useful for treating signals with sharp jumps in volume, where no knee might make compression a little too jarring.

The dog. Tell me about the dog.
Imagine the give in an owner’s arm as they hold a leash, or using one of those ticker leashes with a button the reel the dog back in. This allows the dog just a little more freedom to roam, and the owner a little more leasure in holding the dog back.

Attack
What does it do?
Attack refers to the compressor to reaction time to jumps in volume, usually measured in milliseconds, though occasionally unlabelled. Shorter attack times mean the compressor will be more reactive to changes in volume; the compressor will bring down the volume more quickly and may compress shorter jumps in volume (particularly high-pitched, staticy sounds). Longer attack times will lead to a more gradual introduction of compression to the signal, and may miss jumps in volume entirely at particularly high values.

How might that sound?
Attack and release effectively distinguish compression from distortion, as their shortest values will react to jumps in volume immediately, and the ratio will decide how softly or hard the signal will be clipped. As attack increases in duration, the sound will shift from sharp dips in volume to a smoother, more consistent reduction of volume.

The dog. Tell me about the dog.
Understand that this is where the analogy is imperfect, but consider attack to be the excitability of the dog. A short attack means that the dog will jump at anything, react to even the slightest provocation, and generally call upon the leash to rein her/him in. A longer attack means the dog will be less reactive and eventually just stoic, potentially ignoring things that might draw the attention of other dogs.

Release
What does it do?
Conversely to attack, release describes the time it takes the compressor to return to equilibrium; a sound may jump up in volume, but depending upon release time, the compressor may still be partially engaged by the next time the sound triggers compression. Attack and release work in conjunction with one another, so it is tricky to say if one lends to certain sounds more than other. It is probably best to experiment with extreme values of attack/release, and gradually dial in setting which suit the signal being compressed.

How might that sound?
Where attack might contribute a little more to whether or not a sound will come out distorted and sharp, release contributes a little more to the pumping qualities of sidechain ducking, but the two are relative, and each signal being compressed will react a little differently from the last.

The dog. Tell me about the dog.
If attack roughly compares to a dog’s excitability, it may be suitable to compare release to temprament, or how quickly a dog relaxes after something gets her/his attention. Short release means the dog will be faster to cool down after barking, leaping up, petitioning someone to be pet, while a longer release means the dog will stay at attention for longer. Again, this is where the analogy is not perfect, though by this point in the guide, the parameters should begin to make more sense.

Sidechain Ducking
What is it?
This is often erroneously referred to as sidechaining (which is simply tying control of a parameter to something else), but is quite simple in practice: what if, instead of the compressor reacting to its own track, was guided by the volume of another? Sidechain ducking is exactly that, and is useful for emphasizing certain parts of a mix.

Can that be related to the dog?
Not really, but its actual use is really quite straightforward. In modern drum and bass, sidechain ducking has come to influence a lot of how snare drums feature in songs. Because producers are out to make snares especially punchy and engaging, and also due to the busy place snares occupy in the audio spectrum, modern tracks will often feature a momentary duck of almost all other elements, to allow the snare to leap out at the listener. In fact, the effect is quite controversial, in some producers’ eyes ruining the dynamics of a track, though that is a discussion for another time.

Peak/RMS Volume
What is it?
These terms describe the means by which volume is measured. It is a function of time, with peak referring to quick, continuous measurements of volume, while RMS (or Root Mean Square) refers to a sort of mathematical average to measure sound more gradually. Some compressors use these two measurements to display volume more precisely, while for other compressors the two measurements actually influence how they operate.

Peak is useful for catching/taming random, short leaps in volume which do not sound audibly louder than the rest of the signal, while RMS is better for judging the overall volume of the signal, the loudness as the listener might perceive.

What about… the dog?
There really exists no way to describe this feature in metaphor; they are really just means of quanitfying volume, and may subtly inlfuence how the compressor works.

Expanders/Upwards Compression
What??
These terms actually decribe very different effects, but often end up used interchangeably, and so have been listed together. Imagine a compressor, except key features have been reversed: Expanders generally describe effects which increase volume beyond the threshhold, rather than reduce it- this effect is often used on reverb, or after effects which decrease volume dramatically. Upward compression refers to compressors which react below volume threshhold, rather than above it. Sometimes compressors actually feature both of these deviations. Most producers would be better-served experimenting with them, rather than trying to conceptualize them.

How do those even relate to a dog?
They do not. These are where the metaphor is rendered useless, unless you want to visualize some sort of psychopath chasing after dogs or repelling them somehow.