2

Click here to load reader

Ten Minute Master no.31

Embed Size (px)

Citation preview

Page 1: Ten Minute Master no.31

046 May 2004 MUSIC TECH magazine

T echniques like de-essing, keygating, ducking and frequency-conscious compression can be

incredibly useful ways of controlling unrulyelements in a mix, such as sibilant vocalsand sloppy bass parts, as well as creatingsome innovative sound treatments. Commonto all these techniques is a unique andsometimes misunderstood facet of signalprocessing – the sidechain.

Many signal processors work by‘listening’ to an audio signal and respondingto changes that occur over time. Consider the example of a compressor – it monitors an incoming audio signal and when the signalexceeds a given amplitude threshold itreduces the gain accordingly. Logically, theoutput of the compressor has a directcorrelation to the material it is presented with – as the signal gets louder, the gainreduction increases proportionally. But what would happen if the compressor werelistening to something other than the originalsource audio?

Sidechain processing makes the crucialdifferentiation between the signal beingprocessed (coming in via the audio inputs)and what the processing circuitry is listeningto (coming in via the sidechain inputs). Asidechain input could be a subtly differentversion of the original sound (perhaps withsome mild form of EQ) or a completelydifferent sound, depending on what yourobjectives are.

Frequency consciousIf you’ve ever attached a single-bandcompressor over something like drumoverheads, you’ll understand how problematic

low-frequency sounds can be to a compressorworking with full-spectrum material. Generallyspeaking, low frequencies define the majorityof a track’s sound energy, so rather than theloud snares tickling the compressor’sthreshold, you’ll probably find the bass drumlargely dictating when gain reduction is beingapplied. In short, the compression is biasedtowards low-frequency sounds, as theygenerate most of the threshold peaks, resultingin a sound that seems to pump alongside thepeaks generated by the bass drum. In termsof auditory perception, our ears respond moreto peaks in the mid-range frequencies(around 1kHz), so although the compressionis technically correct, it’s somewhatcontradictory to how we perceive sound.

But what if we could change how thecompressor perceives the sound of thedrums to make it more akin to how our earsoperate? In frequency-conscious compression,a duplicate version of the original sound isEQ’ed and patched into the sidechain inputsof the compressor. The compressor nowlistens to a modified version of the signal,and therefore responds in a different way tothe material presented.

Configuring the EQ has an almost reverselogic to it – by boosting frequencies in thesidechain EQ, the compressor will, in turn,apply more gain reduction, effectively turningdown the equivalent frequency peaks in theoriginal material. Conversely, if you want thecompressor to ignore specific frequencies,these should be removed from the sidechainEQ accordingly. Remember that in configuringthe sidechain EQ, you can be as extreme asyou like – the sidechain input should never beaudible in the end result.

De-stressing de-essingIn essence, de-essing is another example offrequency-conscious compression – just avery application-specific one. In this case,making the compressor respond to excessivesibilance in a vocal track. In de-essing, thesidechain EQ becomes even more important– the compressors must specifically respondto the sibilance and cannot be swayed by anyof the part’s other vocal signal. Exactfrequencies need to be boosted (by up to12dB) to pick out the sibilance (generallyaround the 5-8kHz region, but the actualfrequency varies between vocalists) so makesure you use your ears.

Ideally, an extreme cut below this givenfrequency should also be applied, which isbest done using a high-pass filter, althoughshelving or parametric EQ can also have agood stab at it, given plenty of cut (12dB ormore) to get the attenuation required. Interms of the compressor, most engineers willuse short attack and release times (to get inand out of the sibilance as quickly as possible)with a medium ratio of about 4:1, although

METHOD TEN MINUTE MASTER No31: SIDECHAIN PROCESSING

Inserting an external sidechain input on acompressor removes the main input’s control

over the gain-reduction circuitry.

>

De-essing: note the sharp spike on the EQ’edsidechain (this is heavily EQ’ed sibilance) and the

corresponding dip of the gain reduction across the maininput as this occurs.

>

Key gating: a pad sound is fed tothe main input, while the hi-hat is

fed to the external key input. Looking atthe output, the pad now pulses insympathy with the hi-hat.

>

TEN MINUTE MASTER

No31: Sidechain processingSidechain processing is the key that unlocks a number of advancing signal

processing techniques. Mark Cousins takes a look at the principles involved.

0dB

-6dB

-96dB

0dB

-15dB

-96dB

Main input

Sidechain input

Main input (pad)

Key input (hi-hat)

Output

gain reduction

Output

Input

Key In

Input Output

Sidechain

threshold

Compressor

Gainreduction

>

>

>

>

MTM14.10mm Sidechain 31/3/04 4:01 AM Page 1

Page 2: Ten Minute Master no.31

May 2004 MUSIC TECH magazine 047

higher ratios might be required with moreextreme sibilance.

Nothing beats the flexibility of using anumber of hardware processors for de-essing,but it can be an expensive way of sorting outa vocal if you lose a compressor and EQ unitin the process. Thankfully, there are a numberof dedicated vocal processing units, bothhardware and software, that include pre-configured, pre-patched de-essers.

The key to the gateSo far we’ve only considered sidechains withcompressors, but the use of side-chainingextends much further than this – it’s equallyimportant when working with a gate. In afurther – and possibly confusing – twist, theterminology can change (although the theoryremains the same), with a gate’s sidechainusually being referred to as its ‘key’. Whenset to an internal key, the gate responds tothe signal arriving at the inputs. When set to

an external key the gate now appears to workin much the same way as a sidechainedcompressor, with the gate responding to itskey inputs, rather than its audio inputs. Inreality, these terms (‘keys’ and ‘sidechains’)are interchangeable to some extent, and canbe found on both compressors and gates.

The use of keys on gates tends to be alittle more radical than sidechainedcompressors that use just an EQ’ed versionof the original signal. Many of the acceptedtechniques actually involve using a completelydifferent sound as the external key to thegate. Easily the most clichéd example(possibly because it’s the most fun) is that of‘key gating’. Here, a long, sustained sound(like a pad) is sent to the gate’s input while ashortened, repeating percussive element issent to the gate’s key inputs. The result isthat the longer sound being fed to the inputsappears to pulse, in a way that correspondsto the signal being fed into the key inputs.Trance music has built itself on thistechnique, sending deep synth pads througha gate keyed from a hi-hat pattern, but thepossibilities extend much further than that.

Although less popular nowadays, keygating is still a quick and easy way of tidyingup sloppy kick and bass parts. Here the bassis gated using an external key, fed from thekick drum track that turns down the basswhenever the kick isn’t playing, bringing theattacks more in line with each other. Carefullytweaked, this can produce the impression ofa tighter, more refined rhythm section,although you’ll need to take care inpreserving at least some of the originalcharacteristics of the bass.

EQ’ing the sidechain, or key input (as withcompressors), makes the gate frequency-conscious. Indeed, this is such a usefulfeature (especially when trying to gate close-mic’ed kick and snare drums) thatmany gates incorporate some form of basickey filter controls. The filters are usually quitetame and of the low- and high-pass varieties,so the option to still use external EQ(patched to the key input) is a welcome one.

Ducking to fitAlthough we’ve looked at gates usingcompletely different sources for their keyinputs, the same technique can also beapplied to compressors – an approach bestexemplified by ‘ducking’. Ducking is oftenused in broadcast studios to automaticallyreduce the level of background music inresponse to the presence of a voice-over – anannouncer talking over the title music of a TVprogram, for example.

In this scenario, the music is fed to thecompressor’s input, while a copy of theannouncer’s mic output is sent into thecompressor’s sidechain input. As theannouncer begins to speak, the threshold istriggered and the gain reduction is appliedwith smooth, graduated attack and releasetimes to ensure that the music doesn’t pump

in and out too quickly. Some pro-level gatesalso provide a ducking control (essentiallyreversing the way the gate works in responsesignals above the threshold), which canprovide a more reliable and controllablemeans of ducking broadcast material.

The vocoder story Vocoding is easily the juiciest example ofsidechain processing, and was covered indepth in a Ten Minute Master in Issue 7. Likeall the other sidechain techniques, it usestwo types of input signal: the modulatorsignal (from which the vocoder will analysethe amplitude of the various frequencybands), and the carrier signal (used by thevocoder to re-synthesise its output).

Interestingly, the carrier/modulator themecould be applied to all the examples we’velooked at – in key gating, for example, thepad is the carrier and the hi-hat is themodulator. Many software vocoders require asidechain to work properly – Emagic’sEVOC20 uses the sidechain input as thesource for its modulator (or ‘analysis’, as itcalls it), while the carrier comes from eitherthe audio track or integrated synth section.

Sidechain processing is the bedrock ofadvanced signal-processing techniques. Masterthe possibilities it offers and you will hugelyexpand the your production horizons. MTM

FURTHER INFOMore information•Logic has had proper sidechain processing

since version 5. The webpage below outlinessome of the interesting creative possibilities of plug-ins such as the ES1 and EXS24:www.emagic.de/support/tipsNtricks/

sidechains.php?lang=EN

•There are plenty of interesting articles on the web about compression, some moretechnical than others:www.trinitysoundcompany.com/

compression1.html

Pro Tools had a sidechain facility long before mostother audio sequencing platforms, although

Digidesign refers to it as an ‘external key’.

>

TECH TERMS

SibilanceThe over-emphasis of ‘s’, ‘ch’ and ‘sh’ sounds inthe 5-10kHz range of speech and singing. A poorchoice of microphone and/or over EQ’ing canexaggerate the effect.

ThresholdThe point (defined in decibels) at which acompressor or gate begins applying gain reduction.A compressor applies gain reduction above thethreshold, while a gate applies it below.

Full-spectrumA sound or piece of music that contains acomprehensive range of frequencies across a largeproportion of the audio spectrum. Examplesinclude drum overheads and complete mixes.

Auditory perceptionAuditory perception defines the mechanics oflistening and how our brains identify, interpret and attach meaning to given sounds.

Vocoding is the mother of all sidechain processingtechniques. Emagic’s EVOC20 uses a sidechain input

to form the modulator, or ‘analysis’, signal.

>

If you’re feeling lazy, there are several software plug-ins (like TC Works’ Compressor/De-essor) that

include pre-configured de-essing elements – just set thefrequency and the amount of gain reduction required.

>

MTM14.10mm Sidechain 31/3/04 4:01 AM Page 2