The compressor is an indispensable processor in sound engineering and is used in many different ways. In this section we will describe its most common and codified uses, bearing in mind the possibility of a more "personal" approach and experimentation, so long as this takes place with method and criteria, and not with the "anyhow something cool will come out of it" approach.
Compression of the whole mix: by applying a stereo compression to an entire mix we can obtain a more homogeneous sound, because its sudden volume variations become leveled and each sound is integrated into a single body of sound.
The compression of an entire mix also allows us to limit its dynamic range, and this is done either because the musical genre requests it (for example dance music requires no more than 30 dB, therefore if we were to record a real drum for a dance music track, we'd have to heavily compress it), or for technical reasons (for example, radio-transmission of the music consents a maximum dynamic range of 15 dB, therefore a heavy compression is applied to the tracks before they are sent on air).
Sometimes a light compression is already applied during recording so as to avoid saturation problems.
Obviously what has been said with regards to mixes, applies to submixes too. For example if we have a wind-section with 8 instruments playing, we could choose to sub-mix and compress them separately to obtain a homogeneous wind-section before they get mixed with the rest of the instruments. The same is true with a drum kit recorded with many microphones. The values used in this context are: light compression ratio (2:1) (to not add manipulations that are too evident), low threshold (because we want to bring about an overall reduction of the dynamic range and not only the upper part), medium attack (to react to more or less all the sounds in the mix) long release (up to 2 seconds, this way the compressor remains practically always in action and avoids ugly volume changes).
When a compressor is used in stereo mode, the two sidechain inputs are linked by an appropriate switch in order to have a synchronized function on two channels. In this case the sidechain input controls are unified on the left compressor whilst the right ones get deactivated.
Modification of the signal envelope: in this case we wish to modify the sound of a single instrument. The choice of parameters depends entirely on the kind of input signal and the result we wish to obtain. Let's take a look at two opposite examples. As our first example, let's imagine that we want to increase the sound's attack. We will therefore use a slow attack time to allow the first sound transient to pass through the compressor unaltered. The release time will have to be longer than the duration of the produced sound, in order to allow the compressor to remain activated until the sound has died out. The threshold and the compression ratio depend on the input signal's amplitude and on how heavily we wish to manipulate the sound. The opposite example is a sound we wish to lengthen in time, as for example in order to obtain a very long sustain on an electric guitar note. In this case the attack time shall be set at its minimum, to not modify the initial part of the sound envelope. The threshold shall be set quite high, the release time shall be long and the compression ratio low. This combination of technical choices ensures that when the sound reaches the threshold, it goes beyond it and thus the compressor is activated. However, as a result of its low compression ratio, the sound's amplitude shall remain almost completely unaltered. After this phase the original sound decays but since the release time is long the compressor maintains the sound's amplitude at its initial value for all its duration.