In this fifth and final part of our Beginner’s Guide to Compression we’re going to be looking at three processes which are closely related to compression: limiting, clipping, and saturation. When used at the final stage in processing your tracks, their goal is almost always to increase the volume as transparently as possible.
Saturation Back in the day, the way that tracks were recorded naturally reduced dynamic range in several ways that aren’t present in a digital system: classic mics warmed and flattened the sound, op amps and valves were driven, and of course everything went down to tape, with all its pleasant, saturating properties. The lack of all these elements contributes to what people often perceive as the cold, sterile sound of digital recordings. If you record traditional instruments today, the choice of the right microphone and preamp with a high-quality analog-to-digital convertor can go a long way to removing that sterility; however, this process does little to help those whose tracks come entirely from digital synths and VSTi’s. Therefore, in recent years there has been a continual stream of digital products designed to emulate the desirable saturation effects of the analog recording chain.
The audible effect of saturation is to bring up the volume of the quieter elements whilst levelling off the louder peaks and adding harmonic distortion. The result is an increase in perceived volume and a kind of warm fuzziness. If you can suffer another visual metaphor, I’d say that if the sound without saturation is a regular domestic cat, then the saturated sound is a fat domestic cat, with long hair that has been back-combed. Current wisdom is that we’re still waiting for the definitive go-to tape saturation emulation, but there are still plenty of useable candidates on the market today for gently fuzzing up, fattening and warming your digital sounds. Voxengo’s Analogflux, Virsyn’s tape simulator, and the saturation plug-in from URS are all worth your investigation. VintageWarmer from PSP is also a popular choice, offering a kind of hybrid saturation, single- and multi-band compression, and limiting effect. Many so-called 'maximizers' fall into broadly the same category – try the UAD Precision Maximizer, or Sonnox Inflator. For those able to invest in a hardware solution, the Empirical Labs Fatso, Cranesong HEDD, and Neve Portico 5042 are unlikely to disappoint.
Limiting As we saw in part two, a limiter is simply a compressor with a very high ratio. Many engineers consider a compressor with a ratio of greater than 10:1 to be a limiter, but more prevalent is the brickwall limiter. As the name suggests, the brickwall limiter takes a none-shall-pass approach to sound, and allows no signal at all to pass above the threshold. Therefore, such a limiter effectively has a ratio of infinity:1. A brickwall limiter is especially useful in the digital age, where it is imperative to keep all recorded signals below the 0db maximum, and failure to do so results in clipping and distortion.
Because the limiter acts as a brick wall, it usually has no attack setting – because if it did, and the attack was set to a high value (ie. slow), the initial part of the sound would likely pass above the threshold before the limiter kicked in, and the limiter would therefore have failed in its task. And in fact, this highlights a problem that the successful limiter has to overcome, namely that of being able to react in time to limit the oncoming audio. For this reason, digital limiters usually employ a ‘look-ahead’ feature – that is, they delay the sound slightly so that they can take a quick look at what is coming and so limit it more transparently and effectively. That way you can be sure that no signal is passing above the limiter’s threshold, and since the limiter’s main use is on the stereo buss at final mixdown, the small latency in the signal caused by the look-ahead feature is unimportant.
The release setting of a limiter, however, is very important – so much so that most limiters feature some kind of automatic setting to attempt to give the smoothest possible result. Unlike compression, where the desired goal might not be transparency but a complete reshaping of the sound source, the goal of limiting is almost always a transparent reduction in the peak volume of the sound; ideally, the limiter would do its job without anyone hearing its effect. In order to achieve this end, you probably do not want the limiter causing the pumping sound that the use of a compressor can offer.
That being the case, most modern digital limiters - such as the popular Waves L2, for example - feature very fast release times. After the limiter is triggered by a peak in the source material, it recovers extremely quickly so as to forestall any audible pumping effect. Using this method, the limiter is able to raise the apparent volume of a track quite dramatically. The downside with this approach, however, is the presence of distortion, which is clearly audible on higher settings. (Usually this is unwanted distortion, but like all audio side effects, can also be used as an effect in its own right.)
In my experience, often the best approach – which is also, happily, the easiest and laziest one – is to set the Release to Auto, and let the software do its work. Whereas in many audio applications the use of presets and automatic settings can yield less than stellar results, this is one in which it seems to work most of the time.
So, if transparency is the goal, how much gain reduction can a limiter usefully effect? With a great limiter placed over the stereo buss, on most source material a 2dB reduction can be achieved without wincing at the results; a 4dB reduction can be achieved with just a slight wince; and a 6dB or more reduction needs a few beers before you can really accept it. Even then, you might be dismayed at how (for example) the cracking snare you were so proud of has suddenly disappeared into the mix, a shadow of its former self. And yet, despite this use of a limiter to squeeze more volume from your tracks, you may still find that your final mix is not as loud as most commercial tracks. Is there a way to squeeze a few more dBs out of your mix?
Clipping Despite being told that clipping the audio signal is a very, very bad thing, ironically you may find the sound of a clipped audio signal to sound more transparent than a limiter. Whilst a limiter has been designed to skillfully round off the peaks as transparently as possible without any of the harsh distortion that clipping generates, in most popular music it’s the snare and the kick that are making up the peaks in the audio and thus are what is going to be affected by the clipping – and the fact is that modern drum sounds can take this type of distortion better than any other instrument. Hell - on very harsh, aggressive electronic musical styles, it’s possible you may even prefer the sound after clipping!
The way the top engineers achieve clipping is by driving the input stage of an expensive mastering-grade analog to digital (A/D) converter. Basically your track is played from an analog source into the converter at an overly high volume, and so the peaks get clipped by the converter as it tries to turn the analog signal into a digital one. The argument goes that the high quality components in the expensive converter allow the signal to be clipped with less artifacts than if you were to just drive the same source through the inputs of your cheap sound card.
This does present the home studio recordist with a couple of problems to overcome, since for one, you probably don’t own such a fantastic mastering grade A/D converter, and secondly, since your track is presumably already in digital format, you’d have to ask yourself whether you want to undergo an additional digital-to-analog conversion just so you can slam it back through an A/D converter once again.
Unfortunately, so far there are surprisingly few software emulations of this effect. Internet warriors sometimes sneer at the T-Racks mastering suite from IK Multimedia, but this is one package which does offer a dedicated clipping algorithm, and I have heard it used to good effect on occasion. Another alternative is GClip, a program which is doubly fantastic on account of it being freeware. For that reason alone I recommend you try it. You may even find that this simple little free plug-in gets you the transparent volume boost that you couldn’t get from an expensive limiter.
Summary When used across the stereo buss, a combination of saturation, limiting, and clipping will likely get your tracks to sound as loud as the latest commercial releases. And, having satisfied yourself that you can compete on that level, your next challenge is to choose not to use them - or at least to use them only in moderation. At every stage of signal processing, check the processed version against the unprocessed version for quality, not volume - you should ask yourself if the corresponding increase in volume is really worth the change in the sound of your final mix, and make the appropriate decision.