What is compression?

The best way to understand compression is to think of it as volume

automation. I like to imagine a little man with ninja-like reflexes sitting
with his finger on the volume fader, turning it up and down as he reacts
to the signal either being too loud or too quiet.
So how does a compressor know when a signal is too loud? How does it know
how much it should turn it down by and how does it know when it should turn it
back up?

These factors are all determined by a few controls.

Threshold
The threshold is the level at which the compressor begins to compress. If your
threshold is set to -10db, then the compressor will start to compress the signal
the second goes over -10db. How does it know how much to compress it by?
This is determined by the ratio.

Ratio
Ratio determines how much a compressor will reduce the gain of a signal
when it crosses the threshold. If you have a 2:1 ratio, then for every 2db that
the signal goes over the threshold it will allow 1db through. For example, if our
threshold is -10db and the signal hits -4db then we have gone 6db over the
threshold. With a ratio of 2:1, it will only allow 3db out of the total 6db through.
You will, therefore, see a 3db gain reduction. An infinite ratio will cause the
compressor to act as a limiter, not allowing any signal to cross the threshold.

Knee
You will typically have a choice between hard and soft knee settings. The
knee dictates how smooth the transition is as the compressor moves from un-
compressed to compressed states. Soft is smoother and hard is more abrupt.

Attack
The attack, measured in ms, is the time it takes for the compressor to fully
compress the signal once it crosses the threshold. If you have a longer attack,
say 80ms this will cause the initial transient to cross the threshold
uncompressed. Playing with attack time is usually how we shape transients.
Longer attack time = punchier transient. Short attack times are used to smooth
out signals and catch peaks.
Release
The amount of time it takes for the compressor to deactivate fully once the
signal drops back below the threshold. Having a longer release can create a
pumping effect, or can be used to clean up reverbs and squash down tails.

Makeup Gain
Generally the last parameter you adjust when setting up your compressor. If a
compressor reduces the gain of a signal by 5db, you can use the makeup gain
to turn it back up.

This is essentially just a volume knob at the end of your signal chain. This
might seem counter-intuitive, I’ve had people asking “why would I compress
the signal only to turn it back up?”

To answer this you need to understand that a compressor is changing the

dynamics of a sound, not turning the entire thing down (unless it has a really
long release and instant attack).

If you have a long attack and the transient is now more prominent, but the tail
has been compressed down by 10db, you might want to increase makeup
gain to really push that exaggerated transient up. If you have used the
compressor to smooth out a vocal but the resulting vocal is now 3db too quiet,
you can increase the makeup gain to compensate.

Note that a lot of compressors have an automatic makeup gain toggle. I would
recommend not using it, rather set the final level by ear than have the
compressor constantly adjusting it automatically.

Dynamic Range Compression

Dynamic Range is the difference in volume between the loudest and quietest
parts of a waveform.
Look at this 909 clap. If the transient peak is hitting at -3db and the tail is
sitting around -15db then the dynamic range is 12db. We can increase the
perceived loudness of this clap by reducing the dynamic range, and then
turning the overall gain up of the clap.
How do we reduce the dynamic range? In our example of the clap we can do
this by turning down the initial peak, but leaving the body and tail unaffected,
and then increasing the overall gain of the clap.

We can see the peaks are all happening within the first 30ms of the clap, the
first transient hits right at 0ms so I would set my compressor to 0ms.

Having a compressor set to 0ms can produce pops and clicks due to the
sharp and instant gain reduction, so set your compressors attack to around
2ms lookahead to around 2ms. This will allow it to pre-empt the signal and
therefore reduce pops and clicks.

Release – I would set this to around 15 to 20ms so that the compressor can
fully release by the time it reaches the body/tail of the clap.

Threshold – I set this around -9db to catch only the peaks

Ratio – I set a ratio really high to almost limit the peaks – this resulted in a
gain reduction of around 4db, squashing down the peaks and giving me the
headroom to turn up the overall gain by 4db

Note that the peak volume of the clap has not changed – but it still sounds a
lot louder.

We could also increase the dynamic range, if for example we didn’t like how
loud the reverb tail was. Simply by having a long enough attack so that the
initial transient doesn’t get compressed, and long release so that the body and

Sidechain Compression
In EDM, Sidechain Compression is mostly used as a “ducking” effect. This
effect is most noticeable in Eric Prydz – Call On Me where you can hear the
vocal and instruments pumping around the kick drum “CALL ON MEEEEE-
EEEE-EEEE-EEEE”.

This works by placing a compressor on the vocal channel, enabling the side
chain feature, and selecting the kick as the input source.
The compressor then reacts to the audio signal of the kick drum instead of the
vocal and will compress the vocal down every time the kick hits. You can
control the intensity of this effect using attack, release, and threshold.

You can also use sidechain compression to clean up other elements in a mix,
for example: Sidechaining the lead to the vocal so that when the vocal is
playing the lead is reduced in volume, making more room for the vocal to
stand out.

Shaping Transients
We’ve already touched on transient shaping but this is essentially either
making transients more exaggerated to help them punch through the mix or to
tame them and push them back in the mix.

To exaggerate a transient follow these steps:

Set a long attack, around 20-100ms depending on the sound. Set the release
so that the compressor resets completely before the next transient hits.

Bring your threshold down about 2-5 dB below the peak. Set your ratio to taste
and finally push up the makeup gain to make that transient louder.

To smooth out transients and push them back in the mix: Set a short attack,
0ms with a 2ms lookahead to prevent popping and clicking

Set a really quick release, around 5-20ms depending on how long the
transient is. Bring down the threshold 2-5db below the peak

Set ratio to taste Makeup gain isn’t usually required unless the goal is to make
the sound louder.

Types Of Compressors
Optical Compressors
Optical compressors work by converting the electricity from the audio signal
into light. This light is detected by a sensor in a photosensitive cell. The
brighter the light, the more the compressor reduces the gain. Due to the
nature of this design, it leads to a very smooth release and a quick attack.
These are typically good for legato type sounds like vocals.
The most famous optical compressor or “opto-compressor” is the LA-2A. A
rather expensive compressor, if you want to buy the original hardware unit you
would be looking at spending around £3500. However, there are many plugins
modeled on this compressor that you can pick up for relatively cheap like the
White 2A I mentioned earlier in the article, the CLA-2A by Waves or the VC-
2A by Native Instruments.

Variable-Mu or Delta-Mu Compressors

These compressors are tube based and work by sending voltage to a vacuum
tube which in-turn changes the tubes bias. Due to the reaction time of this
style of compressor, we get a relatively smooth and organic kind of
compression. Another characteristic of this kind of compressor is the ratio
increases dynamically as the voltage increases – so the louder the transient,
the more it gets compressed. These are great as a final touch on a mix,
running an already well-mixed track through a Vari-Mu compressor will help
smooth it out and glue the mix together.

The most well-known Variable-Mu compressor is Manley’s Vari-Mu, if you

don’t have £3000-4000 lying around, try plugin emulations like the Fairchild
670 and 660, or the T-Racks Vintage Compressor.

FET Compressors
FET stands for Field Effect Transistor. These compressors use transistors
instead of Tubes. Due to this design, we get a very quick and snappy
compressor that adds warmth and is good for preserving transients. The first
compressor made in this style was the UAD 1176. The fast attack times were
the main selling point of this compressor. The recommended application for
these compressors would be anything with quick and sharp transients like
your drum bus or an arpeggiator.

The great thing is these compressors are now available in plugin form made
by UAD themselves. Check out the 1176 Classic Limiter Collection by UAD.

VCA Compressors
VCA stands for Voltage Controlled Amplifier. They compress based on a
control voltage and use solid state or integrated circuits. This allows for even
more control over attack and release times and results in a cleaner, more
transparent type of compression. The most well known VCA compressor is the
SSL G-Series Bus Compressor. These compressors are best used for “gluing”
mixes together, hence why it’s called a Bus Compressor.