Compression

Types of Compression

There are two types of compression, both reducing the dynamic range of the signal, but in the real world only downward compression is used.

• Downward Compression Everything above the threshold is turned down.
• Upward Compression Everything below the threshold is turned up

Compression is used to control the dynamic range of a sound. Some sounds can have a huge dynamic range - the quiet parts are very quiet and the loud parts are very loud. This means it is hard to hear both parts well within a mix because either the loud parts will be too loud (in order to bring up the quiet parts) or the quiet parts will be too quiet (in order to make sure the loud parts aren't too loud).

A compressor decreases this dynamic range by reducing the volume of the loud parts so that there is less difference with the quiet parts, unifying the sound and making it easier to work with.

A compressor is effecting the volume of the signal, not the frequency.

Common Controls

• Threshold represents a volume level. When the signal crosses the threshold, the compressor kicks in.
• Ratio is the amount of compression that should be applied when the threshold is crossed. It is in the form of a ratio based on the amount of compression to be applied to the signal relative to the threshold, so 4:1 means that a signal that is 8dB over the threshold will be reduced to 2dB over the threshold. 1:1 means there is no compression. Infinite means that the compressor will perform like a limiter, reducing everything down to the threshold.
• Attack _controls how quickly the compression starts after the t_hreshold is reached. A slower attach allows more of the original signal through. A faster attack allows less.
• Release _controls how quickly the volume returns to uncompressed/unity gain after the volume drops below the _threshold.
• Input Gain controls how much the signal should be amplified before it is compressed.
• Output Gain controls how much the signal should be amplified after it is compressed.
• Knee Settings softens the transfer curve so compression doesn't kick in suddenly at the threshold but is gradually introduced. This provides a more natural sound.
• Adapt Release / Auto Release adapts the release time depending on the input signal so that it is faster for more abrupt sounds and slower for more lengthy sounds. With a slow release, a sudden very loud signal can mean the compressor takes a long time to recover, 'punching a hole' in the sound. With a fast release, more ambient sounds 'pump' sounding unnatural. Rather than finding a compromise this allows the compressor to adapt to both circumstances.
• Auto Makeup Gain attempts to automatically apply gain so that compression doesn't reduce the gain. i

Uses

• Reducing Dynamic range of signal.
• Change the volume ADSR of a sound.

The Detector

Side Chain Compression

By controlling what is supplied to the compressor's detector, the behavior of the detector can be altered. By EQing the input signal, the detector will only look at the EQed signal when deciding whether or not to apply compression. It is also possible to feed a completely unrelated signal into the detector, causing the detector to apply compression to an unrelated signal. For example a drum track can be sidechained in to the detector, causing it to apply compression to a pad when the drums hit. This could be used to get the pad out the way of the drums, or if the drums are not audible, as an effect on the pad.

Uses

• Side chaining drums into a baseline to make it pump.
• Side chaining one track into another so that it gets out of the way at the right moments.
• Side chaining one track into its own delayed signal to make sure they don't overlap (ducking).

Modes (Peak vs RMS)

These modes control how the detector calculates the loudness of the signal it is receiving.

• Using Peak means the detector looks at the loudest part of the signal.
• Using RMS means the detector looks at an average of the signal.
• Because RMS is an average, it will always be less than Peak.

Stereo Linking

Usually both channels are processed simultaneously by a compressor - they are stereo linked. However if the stereo supports stereo-linking the channels can be unlined so that each channel runs through a different detector and has compression applied independently.

Types of Compressor Design

• Feed Forward Design uses a detector processing a side-chained uncompressed signal to apply gain modulation. This is a newer design and is more exact and consitant.
• Feed Back Design uses a detector processing a side-chained compressed signal to apply gain modulation. This is an older design and can feel slightly more natural.

Some compressors can change between the two. Both have a different feel and different applications.

Types of Compressor Circuit

• Tube Compressors are the first compressors made. They use vacuum tubes to control the gain. Sound smooth, silky and fat. Not known for drastic compression. Usually feed-back design.
• FET(Field Effect Transistor) Compressors. Allow very fast attack and release times. Less consistent detection across models so not good paired for stereo compression. Usually feed-back design. More agressive and coloured sound. Great for drastic compression where artifacts are part of the sound.
• Optical Compressors uses changes in light to control the gain. Slower attack times so feel like RMS compression. Able to adapt to the incoming signal (like auto-release). Usually fixed ratio and soft knee. Minimal controls. Usually feed-back design. Transparent musical design.
• VCA(Voltage Controlled Amplifier) Compressors are the most modern compressors. Capable of most neutral and clean compression but can also be very colourful. Very accurate and consistent. Usually feed-forward.

Setting Up A Compressor

Order

Before adjusting the compression settings, the Ratio, Release and Threshold should be set up. By setting Ratio and Threshold to high values it is easier to hear the effect of Attack and Release. By setting the Release to a low value, the effect of the attack is not impaired.

Before

• Ratio 20:1
• Release Less than 20ms
• Threshold Set to give ~20–30dB Compression

The compressor should now be configured based on how the signal sounds. Firstly the shape of the compression is deduced by changing the Attack and Release, then the Ratio and Threshold are reduced according to requirements and taste.

Adjust

• Attack Increase attack from a low value until it sounds right. Try and keep the reduction between 20–30dB so you can clearly hear the effect and adjust Threshold if needed. Adjust the makeup if needed to hear signal.
• Release Increase release until it feels rythmically right. Don't take it so far that too much of the signal is sucked away.
• Ratio Decrease ratio to find balance of size of signal and compression effect.
• Threshold Increase the threshold to reduce the amount of compression. You might want to increase the release slightly as well as less compression should allow for a longer release.
• Makeup Gain Bring the signal back up to its pre-compression volume.

Decision Process

Before applying compression, consider what the current dynamic range of the signal is and what you would like it to be. If you are trying to even out a vocal or instrument for example look at how loud the loudest peak and the quietest peak are and work out the difference in volume. To bring down the loudest peak so that it matches the quietest peak you will need to apply that amount of compression. For example, if the loudest strum of a guitar is -5dB and the quietest is -9dB, you will need to compress the signal by -4dB.

Choosing A Compressor

• Individual drums tend to have similar length hits/tails, so a compressor with a fixed release can be used.
• Bass guitar often has varying length hits and tails, so a compressor with an automatic release helps avoid compromise or pumping. Use of a soft knee can help keep the bass sounding natural.
• Synth basses often have a wide frequency range and include high frequencies. It might be necessary to split the signal into low (below 200Hz) and high(above 200Hz) parts and apply compression on them separately.
• Lead Vocals almost always have a huge dynamic range, so like the bass, something with an automatic release and a soft knee will help.
• Backing Vocals usually want to be layered underneath more dynamic lead vocals, so need to be heavily compressed to iron-out any peaks and leave them consistent with minimised dynamics.

Types of Compression

Serial Compression

Sometimes a single compressor won't be able to achieve everything, for example on a vocal with a broad dynamic range you might want to target the peaks with one compressor, then place another compressor in series to treat the whole signal. It is important to remember that when compression happens in series, the compression ratios are not added but multiplied together, so if the first compressor has a 3:1 ratio and the second a 4:1 ratio, the total amount is 12:1 (3*4).

Parallel Compression

Because compression can often result in a loss of dynamic range (indeed that is often the purpose), parallel compression offers a way to maintain transients alongside the compressed signal. A signal is compressed, then mixed back into the original signal. The original transients are retained from the original signal and the compressed signal (when increased in volume) sits alongside, adding body. In many ways this is like upward compression because the quieter areas have become louder.

Multiband Compression

Multiband compression is the use of multiple compressors in parallel, with each targeting a different frequency band of the original signal. The signals are then recombined. Most effective on a signal with a wide frequency range. However there can be issues with phase-discrepancy so subtlety is key.

Mix Bus Compression

As opposed to adding compression to a finished mix, by placing a compressor over the track whilst mixing it, the decisions made in mixing are informed by the compressed sound. This can make the mix more cohesive. It also controls the level of the mix, with the compressor catching anything that is over-loud and pushing it back down.

Styles

• Gentle Levelling: Slower attack times. Release grooving with the song. Low ration 2:1 or less. Soft knee. Threshold set so compressor is detecting most of the time. Results in around 2dB of compression. Gentle, transparent compression.
• Heavy Compression Mid–fast compression. Release grooving with song, ratio 2:1–4:1. Mid–hard knee. Threshold set so compressor working most of the time, Results in around 4dB of compression. Creates pumping / movement.
• Peak Control Faster attack times. Faster release times, ratio 4:1+. Hard knee. Threshold set so compressor only clamping down on peaks. Ratio 1–4dB. Control the peaks, allowing mix to be pushed harder. Like hard tape compression.

Mid-Side Compression

Frequency Dependent Compression

Target a particular frequency for compression.