This is undoubtedly the most important processor. The compressor works on the dynamic range [Dynamic Range ] of the input signal and reduces its amplitude when it goes beyond a certain limit: this reduction is expressed by a ratio, for example 3:1. This means that when the signal goes over the limit, the part of the signal above this limit gets reduced by 1/3:
In the previous figure we can see how on the left, we have the signal which presents itself at the compressor's input. On the left again, we see the reference-amplitudes measured in dBu and we notice how the signal has a dynamic range of 50dB. The diagram also shows the threshold-point chosen for the compressor's action. On the right-hand diagram we see the result of a 3:1 compression. The part of the signal below the threshold-point has remained unvaried whilst the upper part has been reduced by 1/3 and thus the part of the dynamic range above the threshold, which was 30 dB, has been reduced to 10 dB. The overall dynamic range has therefore been reduced from 50 dB to 30 dB.
Let's now have a detailed look at the compressor's controls:
Threshold: This value is expressed in dB and determines the threshold beyond which the compressor activates itself.
Ratio: Quantifies the reduction in signal-amplitude above the threshold. Some typical ratios are:
1:1 - No compression, the output signal is the same as the input signal.
2:1 - the signal above the threshold is halved. If the signal goes over the threshold by 10 dB its value will be reduced to 5dB above the threshold.
Other values include: 3:1, 4:1 etc. For values higher than 10:1, the compressor behaves practically like a limiter [Limiter ] .
In the following diagram a compression curve of a compressor is shown, for different compression-ratio values:
Compression curve
The diagram shows the amplitude of the output signal related to the input signal. We can see how up to the threshold value the signal's amplitude is the same as that of the input-signal. Beyond the threshold, compression takes place according to the set ratio.
Attack Time: Indicates the time taken by the compressor to be activate itself after the signal has gone beyond its threshold. It is indicated in milliseconds. In the following diagram two situations with respectively short and long attack times, are compared.
Attack times of a compressor
To allow a long attack-time means that the signal has gone beyond the preset threshold, but until the attack time has passed, it won't get compressed. Once the attack time is over, the compressor reduces the signal's amplitude: this results in a highlighting effect of the initial part of sounds. Indeed let's imagine a bass drum whose sound envelope [Sound Envelope ] initially has the form highlighted in green:
Compressor and ADSR envelope
By applying compression, the envelope becomes that indicated in red. This strongly highlights the bass-drums attack, giving it a sharper sound. Two extreme examples of the sound of a bass drum can be found in techno-music and jazz. In techno-music if the sound of the bass drum isn't completely synthetic the sound of the bass drum must be very sharp, dry, aggressive, and therefore a high degree of compression takes place (4:1, for example) with a slow attack time (100 ms, for example). In jazz music the sound of the bass-drum is to be considered effectively almost like the sound of an actual instrument, and therefore has a long tail-sound, almost like a booming. In this case we'll use a lighter compression ratio (2:1, for example) and a very brief attack-time (10 ms) to capture the whole sound envelope. For physical reasons, it is impossible to produce analogue compressors which have a very short, or complete lack of attack-time. This depends on the fact that the circuits physically have a reaction-time at every signal variation. A lack of attack-time can be simulated on a sampled signal and memorized onto a RAM: in this case the compressor is already aware of the whole of the rate of the signal it will manipulate, and it is therefore possible to make elaborations with zero attack-time, although not in real-time.
Release time: this is the time taken by the compressor to return to absence of compression, in other words a 1:1 ratio after the input-signal has gone below its peak. Its purpose is to soften the compressor's activity.
Hold time: After the input-signal's amplitude has returned beneath the threshold, the compressor reduces its action during release time until it reaches 1:1 compression ratio. The hold time allows release time to be delayed after the signal has gone beneath the threshold. It practically keeps the compressor activated for longer.
In the following diagram the actions of a compressor in two phases is demonstrated:
Compressor in action
The following is the pure sound of a bass drum (the one you play with a pedal) manipulated by a compressor which modifies its ADSR envelope
Bass drum [Track 39]
![Effects and signal processors - Bass drum [Track 39]](/images/en/figures/layout/icona_suono.jpg "Effects and signal processors - Bass drum [Track 39]")
Compressed bass drum [Track 40]
To have a better perception of the compressor's effects it is useful to observe its effects on ADSR envelope. In the following figure we see the envelope of a bass drum beat extracted from the previous sound, and thereafter the very same envelope after compression has been applied to it. A comparison between the two figures clearly highlights the compression operation.
Bass drum
Compressed bass drum
Key input
The compressor circuit can be seen as an amplifier controlled by a tension whereby the controlling tension is that of the input signal. If the input signal's tension goes beyond the threshold, the compressor is activated. It is not necessary for the compressor to be controlled by the input-signal's tension- any controlling signal will do. This peculiar trait of compressors allows a whole series of very interesting artifices to take place. Let's see an example: when the bass drum is covered by a bass note played contemporarily, in particular on the even beats (1 and 3 in 4/4 music). This is quite a common situation, because the frequency-content of the two sounds is similar and therefore easily confused. Let's see how we can highlight the sound of the bass-drum in the moment in which it is beaten. Firstly let's compress the bass-drum, as we saw earlier, with a high compression-ratio and a slow attack time, with the intent of emphasizing the bass drum's attack, its "punch". Then let's take another compressor and let's apply it to the bass' signal, using the signal from the bass drum as a sidechain input. This has the effect of lowering the volume of the bass when the bass drum is beaten and therefore the sound of the latter will be the predominant one of the two. After the attack, the compressor goes into its release phase, which means that the bass' volume gently increases: when the bass drum's sound is gone, the compressor ceases its action and the bass returns to its original volume. If we apply a frequency of an LFO [7 ] to a compressor's sidechain input, we create a tremolo effect [Tremolo ] .
We've seen how the shape of a compression curve changes as compression-ratios vary. This kind of curve is called "hard-knee" and presents a rapid variation in gain slope. Another operative modality called "soft knee" has a mellower variation and gives compressors softer functions. The following are the two rates of a compression curve:
Soft and hard knee rates of a compression curve
Compressors acts upon the signal depending on the input tension's rate. There are two operative modalities:
Peak: the compressor responds to signal peaks and therefore measures exactly the amplitude of the input tension.
RMS: the compressor responds to the RMS (Root Mean Square) of the signal, in other words, its effective value, thus it has a softer functioning and less jerky.
This module is capable of bringing about a sub-division of the signal into frequency bands and of operating a different compression on each of the bands. To do this, the module assembles a crossover circuit [The Crossover ] which sub-divides the signal into separate bands. Every output of the crossover is sent to a separate compressor, each of which have independent controls.
Multi-band Compressor
This allows for a lot more refined compressions. Generally high-frequency signals get compressed with rapid attack times and slower release times. This assures that compression follows the input signal's characteristics more precisely.
[7 ] Low Frequency Oscillator is an oscillator capable of generating low-frequency wave-forms (0-10Hz)
Related topics on Wikipedia
Normal dynamic range and compressed dynamic range