A Concise Guide To Compression & Limiting by Paul White

Settings and Characteristics

1 Threshold
2 Ratio
3 Hardknee
4 Attack
5 Autoattack/ Release
6 Holdtime
7 Stereo Link
8 All in the ear
9 Using compressors
10 Side Effects
11 De-Essing

Introduction
Compression and limiting have been covered before, but like the brown mould that you blitz every few months in the bathroom only to watch gradually return, questions on the subject steadily build up again, mere months after we explain the basic principles in an article such as this one! On the one hand, musicians are encouraged to give an enthusiastic and dynamic performance, while on the other, their levels must be controlled to some extent, if we are to create musically acceptable mixes. One tool that is vital in helping us to do this is the compressor, but before looking at how they work, I'd like to outline the types of problems they are designed to solve. While the faders on a mixer can be used to set the overall balance of the voices and instruments that make up a piece of music, short term changes such as the occasional loud guitar note or exuberant vocal scream are less easy to deal with manually.
When I first started recording, compressors were too expensive for home use, so we had no alternative but to 'ride' the faders. Once you've used a compressor to control your levels, however, you come to appreciate that there are certain things it can do that the human engineer is just too slow to manage. For example, unless you've played the track through and memorised exactly where the loud and quiet spots are, you'll always respond too late, because you can't start to move the fader until you hear that something is wrong. A compressor, on the other hand, will be aware of a level problem virtually as soon as it happens. Fortunately, good compressors are now relatively inexpensive, and next to reverb, a compressor is probably the most important studio processor to own—at least for those who work with vocals or a lot of acoustic instruments. For the benefit of those who are still a little unsure as to what a compressor does, it simply reduces the difference between the loudest and quietest parts of a piece of music by automatically turning down the gain when the signal gets past a predetermined level. In this respect, it does a similar job to the human hand on the fader—but it reacts much faster and with greater precision, allowing it to bring excessive level deviations under control almost instantaneously. Unlike the human operator though, the compressor has no feel or intuition; it simply does what you set it up to do, which makes it very important that you understand what all the variable parameters do and how they affect the final sound. In order to react quickly enough, the compressor dispenses with the human ear and instead monitors the signal level by electronic means. A part of the circuit known as the 'side chain' follows the envelope of the signal, usually at the compressor's output, and, uses this to generate a control signal which is fed into the gain control circuit. When the output signal rises past an acceptable level, a control signal is generated and the gain is turned down. Figure 1 shows a simplified block diagram of a typical compressor circuit.

Threshold:
With manual gain riding, the level above which the signal becomes unacceptably loud is determined by the engineer's discretion: if it sounds too loud to him, he turns it down. In the case of a compressor, we have to 'tell' it when to intervene, and this level is known as the Threshold. In a conventional compressor, the Threshold is varied via a knob calibrated in dBs, and a gain reduction meter is usually included so we can see how much the gain is being modified. If the signal level falls short of the threshold, no processing takes place and the gain reduction meter reads OdB. Signals exceeding the Threshold are reduced in level, and the amount of reduction is shown on the meter. This means the signal peaks are no longer as loud as they were, so in order to compensate, a further stage of 'make-up' gain is added after compression, to restore or 'make up' any lost gain.

Ratio:
When the input signal exceeds the Threshold set by the operator, gain reduction is applied, but the actual amount of gain reduction depends on the 'Ratio' setting. You will see the Ratio expressed in the form 4:1 or similar, and the range of a typical Ratio control is variable from 1:1 (no gain reduction all) to infinity:1, which means that the output level is never allowed to rise above the Threshold setting. This latter condition is known as limiting, because the Threshold, in effect, sets a limit which the signal is not allowed to exceed. Ratio is based on dBs, so if a compression ratio of 3:1 is set, an input signal exceeding the Threshold by 3dB will cause only a 1 dB increase in level at the output. In practice, most compressors have sufficient Ratio range to allow them to function as both compressors and limiters, which is why they are sometimes known by both names. The relationship between Threshold and Ratio is shown in Figure 2, but if you're not comfortable with dBs or graphs, all you need to remember is that the larger the Ratio, the more gain reduction is applied to any signal exceeding the Threshold.

Hardknee:
This is not a control or parameter, but rather a characteristic of certain designs of compressor. With a conventional compressor, nothing happens until the signal reaches the Threshold, but as soon as it does, the full quota of gain reduction is thrown at it, as determined by the Ratio control setting. This is known as hard knee compression, because a graph of input gain against output gain will show a clear change in slope (a sharp angle) at the Threshold level, as is evident from Figure 2. Other types of compressor utilise a soft knee characteristic, where the gain reduction is brought in progressively over a range of 1OdB or so. What happens is that when the signal comes within 1OdB or so of the Threshold set by the user, the compressor starts to apply gain reduction, but with a very low Ratio setting, so there's very little effect. As the input level increases, the compression Ratio is automatically increased until at the Threshold level, the Ratio has increased to the amount set by the user on the Ratio control. This results in a gentler degree of control for signals that are hovering around the Threshold point, and the practical outcome is that the signal sounds less obviously processed. This attribute makes soft-knee models popular for processing complete mixes or other sounds that need subtle control. Hard knee compression can sometimes be heard working, and if a lot of gain reduction is being applied, they can sound quite heavy-handed. In some situations, it can make for an interesting sound—take Phil Collins' or Kate Bush's vocal sounds, for example. The dotted curve on the graph in Figure 2 shows a typical soft-knee characteristic.

All in the ear:
You may have noticed, or at least read about, the fact that different makes of compressor sound different. But if all they're really doing is changing level, shouldn't they all sound exactly the same? As we've already learned, part of the reason is related to the shape of the attack and release curves of the compressor, and of course peak sensing will produce different results to RMS, but at least as important is the way in which a compressor distorts the signal. Technically perhaps, the best compressor is one that doesn't add any distortion, but most engineers seem to like the 'warm' sound of the older valve designs which, on paper, are blighted by high distortion levels. The truth is that low levels of distortion have a profound effect on the way in which we perceive sound, which is the principle on which aural exciters work. A very small amount of even-harmonic distortion can tighten up bass sounds, while making the top end seem brighter and cleaner. The best-sounding contemporary compressor designs include valve models with a degree of distortion built in, while others use FETs, which mimic the behaviour of valve circuits. As digital recorders and mixers are introduced into the signal chain, more people are becoming interested in equipment that can put the warmth back into what they perceive as an over-clinical sound.

Using compressors:
One problem newcomers to recording seem to have is deciding where in their system to patch the compressor. A compressor is a processor rather than an effect, so it should be used via an insert point or be patched in-line with a line-level signal. If you have a system without insert points and you want to compress a mic input, you may be able to use your foldback (pre-fade send) in an unconventional way to get around the problem, as shown in Figure 3.
Here's how to do it: Plug the mic into a mixer channel, set the mic gain level as normal, but turn the channel fader completely down. Turn the pre-fade aux send control to around three-quarters up, and do the same with the pre-fade master control, if there is one. Turn the pre-fade send fully down on all the other channels. Now you can take your mic signal (now boosted to line level), from the pre-fade send output, feed it into the compressor and bring it back into another channel of the mixer—this time into the line input. And there you have it: your compressed mic signal. Most engineers will normally add some compression to vocals while recording, and then add more if necessary while mixing. Working this way makes good use of the tape's dynamic range, while helping to prevent signal peaks from overloading the tape machine. It is best to use rather less compression than might ultimately be needed while recording, so that a little more can be added at the mixing stage if required. If too much compression is added at the beginning, there's little you can do to get rid of it afterwards. Similarly, if you have a compressor with a gate built-in, it might be better to leave this off when recording, and only use it while mixing. This will prevent a good take from being wrecked by an inappropriate gate setting. A further benefit of gating during the mix is that the gate will remove any tape hiss, along with the original recorded noise. If a gate is allowed to close too rapidly, it can chop off the ends of wanted sounds that have long decays, especially those with long reverb tails, so most gates (and expanders) fitted to compressors have either a switchable long/short release time, or a proper variable-release time control.

Stereo Link:
When processing stereo signals, it is important that both channels are treated equally, for the stereo image will wander if one channel receives more compression than the other. For example, if a loud sound occurs only in the left channel, then the left channel gain will be reduced, and everything else present in the left channel will also be turned down in the mix. This will result in an apparent movement towards the right channel, which is not undergoing so much gain reduction. The Stereo Link switch of a dual-channel compressor simply forces both channels to work together, based either on an average of the two input signals, or whichever is the highest in level at any one time. Of course, both channels must be set up exactly the same for this to work properly, but that's taken care of by the compressor. When the two channels are switched to stereo, one set of controls usually becomes the master for both channels—though some manufacturers opt for averaging the two channel's control settings, or for reacting to whichever channel's controls are set to the highest value.

Side Effects:
Most of the sound energy in a typical piece of music occupies the low end of the audio spectrum, which is why your VU meters always seem to respond to the bass drum and bass guitar. High frequency sounds tend to be much lower in level and so rarely need compressing, but even so, high-frequency sounds in the mix are still brought down in level whenever the compressor reacts to loud bass sounds. For example, a quiet hi-hat occurring at the same time as a loud bass drum beat will be reduced in level. One technique to reduce the severity of this effect is to set a slightly longer attack time on the compressor, to allow the attack of the hi-hat to get through before the gain reduction occurs. This is only a partial solution, and if heavy compression is applied to a full mix, the overall sound can become dull, as the high-frequency detail is reduced in level. Going back to the subjective effect of subtle harmonic distortion for a moment, some compressor designs make use of harmonic distortion or dynamic equalisation to provide an increase in high-frequency level whenever heavy compression is taking place. This helps offset the dulling of high-frequency detail, and can make a great subjective difference, but it isn't a perfect solution. More elaborate compressors have been designed which split the signal into two or more frequency bands and compress these separately. This neatly avoids the bass end causing the high end to be needlessly compressed, but it can introduce other problems related to phase, unless the design is extremely well thought-out.

De-Essing:
Another side chain-related process is the de-essing of sibilant vocal sounds. Sibilance is sometimes evident when people pronounce the letters 's' or 't', and is really a high-pitched whistling caused by air passing around the teeth. If a parametric equaliser is inserted into the side-chain signal path of a compressor and tuned to boost the offending frequency, the compressor will apply more gain reduction when sibilance is present than at other times. Most sibilance occurs in the 5 to 1OkHz region of the audio spectrum, so if the equaliser is tuned to this frequency range and set to give around lOdB of boost, then in the selected frequency range, compression will occur 1OdB before it does in the rest of the audio spectrum. The equaliser should be set up by listening to the equaliser output, and then tuning the frequency control until the sibilant part of the input signal is strongest. Figure 4 shows how a compressor and equaliser may be used as a de-esser. Some compressors have a built-in sweep equaliser, to allow them to double as de-essers without the need for an external parametric equaliser.

I should stress that these are just to get you started—the ideal settings vary from compressor to compressor, which is why I come up with slightly different figures every time I write on the subject. The more gain reduction is used, the higher the level of background noise, so never use more gain reduction than is necessary. Virtually all recorded pop music has a deliberately restricted dynamic range, to make it sound loud and powerful when played over the radio. The more a signal is compressed, the higher its average energy level. In addition to compressing the individual tracks during recording or mixing, the engineer may well have applied further compression to the overall mix. This can be very effective, but don't choke the life out of a mix by over-compressing it either. When it comes to individual tracks, it is pretty much routine to compress vocals, bass guitars, acoustic guitars and occasionally electric guitars, though overdriven guitar sounds tend to be self compressing anyway! The most important of these to get right is the lead vocal, because even modest dips in level can make the Iyrics difficult to hear over the backing. Sequenced instruments are less likely to need compression, because you can control the dynamics by manipulating the MIDI data in the sequencer. My own rule is to avoid compression (or any other form of treatment) unless it's absolutely necessary. Even with vocals, if somebody gives me a perfectly controlled vocal take, I wouldn't want to compress it just because compressing vocals is the done thing. Compression is a very valuable studio tool, but like all tools, it is just a means to an end—not an end in itself.

King_Odin schreef op 15-04-2006 @ 11:40
Wat voor melding krijg je dan? Wat lukt er niet?

Jos de Boss schreef op 15-04-2006 @ 12:02
(...)

Ik zal zo even kijken wat er precies stond, maar ik kreeg iig in fruityloops te lezen dat er een .Dll bestand miste ofzo en dat ik het maar overnieuw moest instaleren alleen toen ik dat deed bleef het hetzelfde, ik heb het een paar keer geprobeerd.

Noedels schreef op 23-04-2006 @ 15:03
Of je knipt zelf

Unexplained schreef op 04-04-2006 @ 10:49
(...)
Nah, drumkits zijn wel leuk en aardig, maar haal mijn snares meestal van vinyl af. Soulplaten, of soms gewoon hiphopplaten. Misschien kan ik nog wel wat regelen met een gast, waar ik normaal bij bestel, of ik die lijst publiekelijk hier mag plaatsen.

Trouwens, mocht je dat "snares vanaf vinyl afhalen" verhaaltje leuk vinden, een gouden tip van ome X.

Numark TTUSB, rond de 200 euro, bij Feedback en consorten. Draaitafel met USB aansluiting.

(...)
(...)

Komik_Sanz schreef op 13-05-2006 @ 21:02
(...)

Word to my man Noedels.
Shit is wack om de drums van bijv. 9th Wonder te gebruiken.

Maar goed, doe wat jullie niet laten kunnen.