″A connecting principle,
Linked to the invisible
Almost imperceptible
Something inexpressible.
Science insusceptible
Logic so inflexible
Causally connectible
Yet nothing is invincible.″
— ″SYNCHRONICITY I,″ THE POLICE, SYNCHRONICITY (A&M RECORDS, 1983)
The typical recording studio has several — sometimes upwards of 100 — faders. The simple fader, nothing more than an adjustable gain stage, is essential to music production.
There are times when a fader does not satisfy an engineer′s need for gain control. Music signals are rarely consistent in level. Every crack of the snare, syllable of the vocal, and strum of the guitar produces a signal that surges up and recedes down in amplitude. Figure 6.1 shows the changing amplitude during about one bar of music. Signals like this one must fit within the amplitude constraints of the entire audio chain (see the discussion of ″Dynamic Range″ in Chapter 1) without damage: the microphone, microphone preamp, console, outboard gear, multitrack recorder, two-track master recorder, power amp, and loudspeakers. The highest peak must get through these devices without clipping, while the detail of the lowest, nearly silent bits of music must pass through without being swamped by noise. When the recording engineer aims for 0 VU on the meters, they are targeting an amplitude that just avoids distortion when the music peaks, yet is not lost in noise when the signal relaxes.
With agile fingers and intense concentration, one could constantly adjust the faders, in reaction to every snare hit, vocal phrase, and guitar flourish. Better yet, let a machine do it. To help narrow the sometimes extremely wide dynamic range of typical audio signals so that they better fit within the amplitude limits imposed by the studio, engineers reach for a compressor — a sort of automatic, semi-intelligent fader. Less intuitively, perhaps, the compressor is also the basis for a wealth of other effects having very little to do with the control of dynamics.
Figure 6.1 The range of amplitude in two bars of music.
This chapter reviews the common parameters found on compressors/limiters, summarizes the core technologies used, and studies in depth the various production techniques — both practical and creative — built on this device.
What is the compressor′s task? Quite simply, when a signal gets too loud, it turns it down. What counts as too loud? How much should it turn it down? How quickly? For how long? Some compressor controls are needed.
A line is drawn separating too loud from not too loud using the threshold control. The threshold setting on the compressor sets the amplitude above which compression is to occur, and below which compression is to stop. As long as the signal remains below this threshold, the compressor is not triggered into action. When the signal exceeds the threshold however, the compressor begins to attenuate the signal, like a fader automatically pulled down. Once the compressor is attenuating a signal, the signal must fall back below this threshold before the compressor will stop attenuating and return to unity gain.
When the audio is above this threshold, the compressor begins to attenuate. The amount of compression is primarily determined by the ratio setting. Mathematically, the ratio compares the amount of the input signal above the threshold to the amount of the attenuated output above the threshold. For example, a 4 : 1 ratio describes a situation in which the input level above the threshold is to be four times higher than the output above the threshold: 4 dB above threshold in becomes 1 dB above threshold out, and 8 dB above threshold in becomes 2 dB above threshold out. A ratio of X : 1 sets the compressor so that the input must exceed the threshold by X dB in order for the output to achieve a level just 1 dB above the threshold.
It is important to note that the ratio only applies to the portion of the signal above the threshold. When the input is below the threshold, the compressor is not applying this ratio to the signal. It is in the business of not compressing signals below the threshold. Only that part of the input that exceeds the user-defined threshold is multiplied by the compression ratio.
It is worth making the distinction now between a compressor and a limiter. The websites, catalogs, and showrooms selling compressors often tout them as compressors/limiters. This is basic marketing; one device seems to have two functions. Yes, dear reader, that is two for the price of one.
Most of this chapter is dedicated to discussing the production potential of the compressor/limiter. Readers of this chapter will find it easy to tap into more than a dozen different effects, all created by this one effects device. The sonic output is very much a function of all of the settings dialed in by the engineer. The compressor/limiter in the hands of a talented engineer is much, much more than two effects for the price of one.
The defining parameter separating compression from limiting is the ratio setting. A ratio below 10 : 1 indicates compression. A ratio above 10 : 1 makes the effect limiting. It is simply a matter of degree. For very tight control of the amplitude of a signal, limit the dynamic range with a high ratio. For less abrupt modification of the amplitude of a signal, gently compress the dynamic range with a low ratio.
As the principle technology for compression and limiting is the same, most compressors are also limiters, and most limiters are also compressors. Therefore, most faceplates and brochures declare the device both: a compressor/limiter.
It is not inappropriate to think of the compressor as a fader, backed by some analysis and intelligence and controlled by a machine. If you have seen moving fader automation on a mixing console, you have a good visual analogy for what goes on inside a compressor. A machine is pulling down and pushing up a virtual fader in reaction to the amplitude of the signal.
The speed with which the signal is attenuated is a function of the attack setting. Attack describes how quickly the compressor can fully kick-in after the threshold has been exceeded.
The importance of attack time is not limited to that transition up through threshold. In fact, the attack character of the compressor is in play for any signal at or above the threshold and increasing in level.
Recall that with a ratio of 4 : 1, an input signal 4 dB above the threshold will be attenuated by 3 dB so that it is just 1 dB above the threshold on output. As the input signal increases to an amplitude that is 8 dB above the threshold, attenuation must increase from 3 dB to 6 dB in order to achieve the target of 2 dB above the threshold output. Attack time governs this further reduction in gain, even as the signal remains above the threshold.
Attack describes the speed of the imagined fader as it turns down the signal whenever attenuation is required. Fast attack times will enable the compressor to react promptly, while slow attack times are more lethargic.
The speed of the imagined fader within the machine as it moves back up toward unity gain is determined by the release setting. When the amplitude of a compressed signal above the threshold starts to head back down in level, for example, during the decay of the snare or at the end of the note, the compressor must begin uncompressing. Where 10 dB of attenuation might have been called for an instant ago, maybe only 8 dB of attenuation is expected now (based on applying the ratio to the now lower portion of the signal above the threshold). The release time setting governs the speed of this reduction in attenuation. That is, the release parameter sets the speed that an attenuating compressor can turn the gain back up toward unity. When the amplitude of the music returns to a level that is below the threshold, the compressor must stop compressing. This same parameter — release — still applies, setting the speed of the compressor as it returns to zero gain change.
Finally, because the compressor attenuates signals based on the four parameters mentioned above it is often desirable to turn the signal back up by a fixed amount, using make-up gain. Managing the dynamic range of the entire signal chain is made easier if one can insert and remove a compressor without a significant change in average overall levels. In the course of a recording session, it is often desirable to compare the compressed and uncompressed signals in order to verify that the effect envisioned is being achieved, without unwanted side effects. Hopefully the signal is in fact better, not worse, for compression. Such comparisons, with and without the effect, benefit from level matching so that the engineer may react to the overall quality of the signal without being distracted by loudness differences. It is common, therefore, to turn the output level up a bit overall.
It is difficult, in the end, to manufacture and then sell a device that makes things quieter. In addition, recording engineers will find it difficult to keep band members happy by making things softer. So a final bump in gain, called make-up gain, is the parameter that raises the overall level of the signal by a single fixed amount.
These five parameters — threshold, ratio, attack, release, and make-up gain — enable the device to carefully monitor and make fine adjustments to the amplitude of a signal automatically. The engineer is then freed to concentrate more on other things: Is the guitar in tune? Is the coffee strong enough?
While these five parameters are always at work, they are not always on the faceplate of the device; that is, they are not always user adjustable. There are compressors at all price points that leave off some of these controls. For example, attack and release may not be user adjusted. They are determined by the manufacturer. The particular attack and release characteristics are an unadjustable part of that compressor′s sound.
Other compressors offer full adjustability of all the parameters yet also offer presets. The presets reflect someone else′s careful tweaking of the parameters for a satisfying sound in specific applications. Certain presets might be recommended for bass, others for drums.
Sometimes the presets simulate the ratio, attack, and release characteristics of other, vintage, collectible, famous sorts of compressors. The reader is encouraged to spend some time with the fully-adjustable types of compressors for exploring and ear training. But do not hesitate to reach for those compressors with only a few knobs on the box during a session. A lack of user-adjustable parameters does not mean a lack of sound quality. They can often get the job done quickly, and with terrific sonic results.
Compression is used to create a variety of effects, detailed below. These particular five parameters are a curious collection of controls: threshold, ratio, attack, release, and make-up gain. It is essential to understand their use in shaping the compression effect, but be forewarned: It is an acquired skill. The user interface of the compressor is not exactly intuitive. Fine tuning the effect during the chaos of a recording session is not at first rewarding. These parameters reflect the adjustable bits of the circuit or algorithm employed by the compressor. They are not necessarily the controls all engineers ideally wish they had. Throughout the early part of any audio career, as more and more sophisticated approaches to multitrack production are obtained, an audio engineer will often be frustrated by these compression parameters. At times they provide only indirect control of the effect one wishes to create. Other times, these controls get in the way, actually interfering with the fine-tuning process. A new user interface is needed, offering a set of controls that has more to do with how compressors are used, and less to do with how they work. So far, no such user interface is in the offing. It does not hurt to hope.
If the compressor were invented today, it would have some hyped-up, one-word-with-two-capital-letters sort of name like PowerFader™, and it would have a website. The humble compressor offers engineers a handy way to precisely control and manipulate the dynamics of the signals recorded.
The signal flow through a compressor (Figure 6.2) consists of an audio path through a gain stage, with a level detector circuit controlling the level of the gain change component. The level detector receives the same signal as the gain change element in most applications. A side chain input makes it possible to feed a modified or entirely different signal into the level detector. This enables a great range of signal-processing effects, as discussed later in this chapter. The parameter settings (threshold, ratio, attack, and release) directly influence the sound of the compressor, but the type of gain change device employed in the design of the compressor makes a significant sonic contribution as well.
Figure 6.2 Signal flow through a compressor.
Tube-based (valve-based) compressors were, in the early days of audio, the only choice for machine-modifying gain. The level detector circuit sends a voltage to the tube that directly drives the gain of the tube. Fairchild and Altec compressors were early devices built on this principle. Today, Manley makes a state-of-the-art update of this approach. The reader is likely very familiar with the often desirable qualities of tube-based components. The sometimes musical and appealing distortion introduced by a tube-based gain stage is part of a tube compressor′s sound (see Chapter 4). In addition, there is the reaction time of the tube as its gain is changed by the level detector circuit. Any decrease or increase in gain is not instantaneous, giving tube-based compressors their own unique attack and release time properties. Discussed below, the rate at which the gain is increased and decreased has a major influence on the overall quality of the compression effect.
Optical compressors use the interaction between a light source and a light-sensitive resistor to influence gain. The level detector circuit essentially illuminates some sort of light (an incandescent bulb, a light-emitting diode (LED), an electroluminescent panel, etc.) whose light shines on a photovoltaic cell. The amplitude of the signal through the compressor is driven by the resistance of the photo cell in reaction to the amount of light shining upon it. The entire LA series of compressors (LA-2A, LA-3A, and LA-4A) formerly by Urei and now made by Universal Audio, as well as some Joe Meek compressors, among others, leverage this principle.
Switch on a light, and there is a time element to how long it takes the light to achieve full illumination. Any good commuter has noticed the difference between light-emitting diodes and incandescent bulbs in the stoplights and taillights around them. LEDs snap on and off all but instantly. Incandescent bulbs turn on more slowly, and linger briefly with a little glow even after they are turned off. Luminescent panels exhibit a memory effect where, if they have been bright for a period of time, they remain illuminated even after the illuminating voltage has been removed.
The particular type of photovoltaic cell similarly contributes to the compressor′s attack and release characteristics. A compressor′s attack and release characteristics are ultimately determined by the complex interaction of the light source and light receiver chosen by the equipment designers.
Solid-state compressors might leverage a field effect transistor (FET) or other transistor-based voltage-controlled amplifier (VCA) to change the gain of the compressor. The Urei (now Universal Audio) 1176LN compressor is a particularly famous use of an FET gain stage. The dbx compressors (old and new) take advantage of high-quality audio VCAs for compression. Using transistors, the detector circuit is empowered to drive the gain via a control voltage.
Compressors in the digital domain are liberated from the constraints of real-time analog circuit components. The qualities of digital compression are as varied as the companies that design them. When audio becomes a string of numbers, digital compressors become a bank of calculations.
One might be tempted to conclude — numbers being numbers — that all digital compressors sound alike. This simply is not the case. Humans write the code that governs the behavior of a digital compressor. There is no right answer, no single solution to compression.
A digital compressor represents one algorithm for achieving compression through calculations, an algorithm that is the result of countless decisions, trade-offs, and moments of inspiration by the creative software engineers who write the code.
Software might be written with the expressed goal of trying to emulate old analog compressors. With so many desirable analog compressors, it makes sense to attempt to capture some qualities of tube, optical, or solid-state compressors digitally. Another valid approach is to create entirely new algorithms that essentially reinvent the compressor-leveraging opportunities not achievable through analog topologies.
6.3 Nominal Application: Dynamic Range Reduction
The compressor is used to create a range of effects, each pursuing different goals artistically, through different means technically. The most intuitive use of compression is to reduce, sometimes slightly other times radically, the audio dynamic range of a signal.
When singers get confident and excited, they may start to sing the choruses very loud — louder than during all the other takes in rehearsal. Uh oh. If this great performance overloads the analog-to-digital converters or the analog tape machine, the track is unusable. Without some amplitude protection, a killer take is easily lost to distortion. Be ready for this with some gentle (around 4 : 1 or less) compression across the vocal. In this way, the equipment will have no trouble accommodating the adrenaline-induced increase in amplitude that comes from musicians when they are ″in the zone.″
During the course of a song, some snare hits are harder than others. The slamming that goes on during the chorus might be substantially louder than the delicate, ghost-note-filled snare work of the bridge. A limiter is employed to attenuate the extreme peaks and prevent unwanted distortion. A limiter is nothing more than a compressor taken out to rather extreme settings: The threshold is high so that it only affects the peaks, leaving the rest of the music untouched; the ratio is high, greater than 10 : 1, so that any signal that breaks above the threshold is severely attenuated; and the attack is very fast so that nothing gets through, even briefly, without limiting.
Figure 6.3 gives an example of peak limiting, the sort of processing used to prevent distortion and protect equipment. Fitting a signal on tape without overloading or broadcasting a signal without overmodulating requires that the signal never exceed some specified amplitude. Limiters are inserted to ensure these amplitude limits are honored. In live sound, exceeding the amplitude capability of the sound reinforcement system can cause feedback, damage loudspeakers, and turn happy crowds into hostile ones. Limiters offer the solution again. They guard the equipment and listeners downstream by confining the signal amplitude to safer levels.
Figure 6.3 Peak limiting.
Analog-to-digital converters, radio transmitters, and tape recorders can be particularly unforgiving of being overdriven. Compressors and limiters are a regular process to add to the signal chain, giving the engineer a protective buffer before troublesome distortion ruins the music.
Frequently music recordings must consciously fight the noise floor of the storage medium (e.g., analog tape) or the noise floor of the listening environment (e.g., an automobile). Compression offers much needed help, narrowing the higher amplitude portions of a signal to a level that is a little nearer the lower amplitude parts. specifically, with the high-amplitude parts attenuated some, the engineer can turn the overall level of the signal up some without subsequent risk of overload distortion. This makes the softest parts of the performance a little louder, lifting them up out of the noise floor.
Some engineering finesse is required, as this application of compression generally does not intend to change the musical dynamics of the signal. The dynamics intended by the composer and the performer should not be undermined. Overcoming noise with compression requires the audio dynamics to be narrowed without harm to the musicality of the piece.
6.3.3 INCREASE PERCEIVED LOUDNESS
Compression is inserted into perfectly fine tracks in order to make them louder. A handy side effect of compressing — reducing the overall dynamic range of the signal — is that now it can be turned up. That is, one can make the track louder as a whole if the more extreme swings in amplitude have been compressed. Figure 6.3 demonstrates this sort of compression. It is counterintuitive at first, but this gain reduction device is used to make a track louder.
With the peaks attenuated, the overall signal can then be raised in level (see the lower part of Figure 6.3). The result, when desired, is an increase in the overall loudness of the signal. The average amplitude is higher, the area under the amplitude curve is increased, and the perceived loudness is raised.
There is no doubt that this effect is overused in contemporary pop-music productions at the beginning of the twenty-first century. The reader is encouraged to use mature musical judgment to find the appropriate amount of peak limiting, creating productions that bear repeated listening by passionate fans without fatigue, yet sound exciting and competitive in the market.
Maximizing loudness is often taken to radical extremes where mixes are absolutely crushed (i.e., really compressed; see also squashed, smooshed, squished, thwacked, etc.) by compression so that the apparent loudness of the song exceeds the loudness of all the other songs on the radio dial or shuffled playlist. Selling music recordings is a competitive business. Loudness does seem to help music sales figures, at least in the short term. And so it goes. Often the music suffers in this commitment to loudness and hope for sales. Artist, producer, and engineer must make this trade-off carefully: long-term musical value versus the short-term thrill of loudness. One must not forget that the consumer has the ability to adjust the level of each and every song they hear by operating the volume knob on their playback system. If they want the song louder, than can make it so. Even in small measures, a little bit of gentle compression buys the tune a little bit of loudness.
6.3.4 IMPROVE INTELLIGIBILITY AND ARTICULATION
There is real artistry and technique in singing and speaking for recording and broadcast. That singers need vocal lessons and practice is accepted. Voiceover artists do too. Part of what is studied and rehearsed is diction. It is essential in almost all styles of music and forms of speaking that each and every word be plainly understood. The intelligibility of words, spoken or sung, depends in part on the ability to accurately hear consonants. Vocal artists are trained to control their vowels and make distinct each and every consonant. Pronouncing a word, the letter B must not be confused with the letter D; N′s and M′s must be differentiated.
Imagine looking at a meter on the mixing console while a singer sings ″bog″ and ″dog.″ The meters would reveal no visible difference between these two words. Imagine looking at the waveform of each word on the screen of your digital audio workstation. They appear, to the naked eye, incredibly similar. There is precious little difference between these two words, differences that aren′t easily measured or confirmed by any device available in the typical recording studio — except our hearing. The incredibly subtle difference is extracted from the signal our ears pick up through the work done by our hearing physiology and the analysis performed by our brain.
Add in hiss, hum, static, and other noises, and distinguishing ″bog″ from ″dog″ becomes more difficult. Add drums, guitars, and didgeridoo to the mix, and the intelligibility is further threatened. Add delay, reverb, and other effects, and hearing detail in the vocal may become more challenging still.
Compression can help. Adjusting the gain of the vocal, syllable by syllable and word by word, the engineer is able to make sure that no important part of a vocal performance gets swallowed by competing sounds. Set the threshold low enough that it is below the average amplitude of the vocal. Reach for a fast attack time and medium to fast release time, and raise the ratio as needed, possibly as high as 10 : 1. The effect is not usually transparent. The vocal is clearly being processed. Done well, though, it can be worth it. Intelligibility rises.
Compression is used often, particularly on the dialog portion of film and television soundtracks, to tame the loudness inconsistencies of the vocal performance so that no portion of the track is difficult to hear. Every syllable of every word is tamed in amplitude by the compressor so that the intellectual meaning of every syllable of every word is communicated clearly.
The articulation of each and every note by a melodic instrument may also carry significant musical value, and yet be drowned out by other elements of a crowded mix. The intelligibility of a horn line or a guitar part can be every bit as important as the intelligibility of the lyrics. Many styles of music, from jazz to progressive rock, make their artistic living based on communicating the exact phrases — the timing and the pitch — of sometimes several instruments at once. Careful control of dynamics through compression may be, at times, a required part of the production.
As with so many effects, tread carefully. This compressor move alters the musical dynamics of the track processed. If the sax player wanted some notes to be softer, with a more impressionistic and expressively vague form of articulation, this kind of compressor effect will be unwanted. Apply only as needed, where artistically appropriate.
When the guitarist gets nervous, she leans to and fro on the guitar stool, leaving the engineer with the challenge of a static microphone aimed at a moving target. A compelling performer, nervous in the studio, still deserves to be recorded. Without the constant gain riding of a compressor, listeners will hear the guitarist moving on and off the microphone. A little gentle compression might just coax a usable recording out of an inexperienced studio performer.
Beware of the vintage guitar. When the bass player pulls out that wonderful, old, collectible, valuable, sweet-sounding, could sure use a little cleaning up, are those the original strings, could not stay in tune for eight bars if you paid it, gorgeous beast of an instrument, we can be sure that — even in the hands of a master — some strings are consistently a little louder than other strings. The instrument is not balanced, a sad fact that might be revealed by abrupt level changes in the bass line being played. Of course, one solution is compression. Careful, note-by-note, precision amplitude adjustments must be made to the signal or the very foundation of the song becomes shaky.
The equal loudness contours (see Chapter 3) highlight the human sensitivity to bass levels. When full-bandwidth signals increase in level, the perceptual result at low frequencies is exaggerated — the lows sound even louder. The perceptually volatile low-frequency portion of our music productions requires particular focus on levels. On bass, audio engineers almost always need the careful level adjustments of gentle compression.
When a drummer′s foot gets (understandably) tired during what may be several takes of the same song, the kick drum performance can become a little ragged. Some individual kicks are noticeably louder than others. Use compression to make the performance more consistent.
It is important to note that smoothing any performance with compression is a signal processing ″fix″ that would never be needed in an ideal world. That is, it is always best to fix these issues at the time of performance, if at all possible. Teach the singer microphone techniques in which they back off the microphone for louder parts of the performance, and lean in slightly on the quieter portions. Find a way to get the guitarist to remain in a relatively steady position by the microphones. After all, the quality, not just the level of the sound, is changed when the acoustic guitar moves toward and away from the microphone.
Find a bass player who is so sensitive to their instrument and performance that they adjust note by note, string by string, for level differences in their playing technique. Great players can create a performance with musically-expressive dynamics across the entire range of their instrument.
If the drummer is getting tired, coordinate with the producer to have strategically-timed breaks so that a consistent performance is recorded to the multitrack and need not be chased with smoothing compression. There is no doubt that a kick drum hit hard has a different spectral content than a kick drum hit softly. Having a compressor match levels of various kicks will not lead to a believable timbre.
When the answering machine was invented, its intended purpose was to answer the phone and take messages while we were out. But the day after the first one was sold, the answering machine took on a new, more important role: call screening. The most common message on these devices is something like, ″It′s me. Pick up. Pick up!″ The use of a device in ways not originally intended occurs all too often, and the compressor offers a case in point. While dynamic range reduction and peak limiting are effective, intended uses for the device, recording engineers use them for other, less obvious, more creative reasons as well.
6.4.1 ALTERING AMPLITUDE ENVELOPE
Compression can be used to change the amplitude envelope of the sound. The envelope describes the ″shape″ of the sound, how gradually or abruptly the sound begins and ends, and what happens in between. Envelope is a step back from the fine-level, cycle-by-cycle detail of the waveform, looking instead at the more general signal amplitude fluctuations as in Figure 6.3. Drums, for example, have a sharp attack and nearly instant decay. That is, the envelope resembles a spike or impulse. Synth pads, on the other hand, might ooze in and out of the mix, a gentle envelope on both the attack and decay side. Piano offers a combination of the two. Its unique envelope begins with a distinct, sharp attack and rings through a gently changing, slowly decaying sustain. All instruments offer their own unique envelope.
Consider the sonic differences among several instruments even as they play the same pitch, say A440: piano, trumpet, voice, guitar, violin, and didgeridoo. There are obvious differences in the spectral content of these instruments; they have a different tone, even as they play the same note. But at least as important, each of these instruments begins and ends the note with its own characteristic envelope — its amplitude signature.
The compressor is the tool used to modify the envelope of a sound. A low threshold, medium attack, high-ratio setting can sharpen the attack. The sound begins at an amplitude above threshold (set low). An instant later (medium attack), the compressor leaps into action and yanks the amplitude of the signal down (high ratio). Such compression audibly alters the shape of the beginning of the sound, giving it a more pronounced attack.
This approach can be applied to almost any track. A good starting point for this sort of work is a snare drum sound. It is demonstrated in Figure 6.4. Be sure the attack is not too fast or the compressor is likely to remove the sharpness of the snare entirely. Set the ratio to at least 4 : 1, and gradually pull the threshold down. This type of compression has the effect of morphing a spike onto the front of the snare sound. Musical judgment is required to make sure the click of the sharper attack fits with the remaining ring of the snare. Adjusting the attack time while trading off a low threshold with a high ratio offers the engineer precise control over the shape of the more aggressive attack. This compressor effect works well on any percussive sound from congas to pianos. It can also sharpen the amplitude onset of less percussive tracks, from the saxophone to the electric bass. Done well, this creates a more exciting sound that finds it place in a crowded mix more easily.
Figure 6.4 Compress to sharpen attack.
Another unusual effect can be created using the compressor′s release parameter. A fast release pulls up the amplitude of the sound even as it decays. This is shown in the snare example of Figure 6.5. Notice the raised amplitude and increased duration in the decay portion of the waveform. Dial in a fast enough release time, and the compressor can raise the volume of the sound (i.e., uncompress) almost as quickly as it decays. This increases the audible duration of the snare sound, making it easier to perceive each drum hit in and among the distorted guitars and ear-tingling reverberation (see ″Masking Reduction″ in Chapter 3). Altering the decay of instruments can be taken to radical extremes. Applied to piano, guitar, and cymbals, the instruments can be coaxed into a nearly infinite sustain, converting them into bell or chimelike instruments in character, while still retaining the unmistakable sound of the original instrument. File under ″Special Effects.″ It can be just the sort of unnatural effect a pop tune needs to get noticed.
Figure 6.5 Compress to lengthen sustain (attack remains sharp).
A coordinated adjustment of compressor parameters (threshold, ratio, attack, and release) enables the audio engineer to manipulate, however indirectly, the shape of the amplitude envelope toward still other goals.
Consider some extreme (i.e., high ratio, low threshold) compression with a very fast release time. If the compressor pulls down the peaks of the waveform and then quickly releases the signal as it decays, listeners may be able to hear parts of the sound that were previously inaudible. Through fast-release compression, the engineer effectively turns up the later parts of the sound, revealing more of the decay of a snare, the expressive breaths between the words of a vocal, the ambience of the room in between drum hits, the delicate detail at the end of a sax note, and so on. Clues about space, emotion, and performance intensity can be brought forward through compression for a powerful musical result.
Pop music standards push productions to have bright, airy, in-your-face, exciting vocal tracks. This vocal sound must rise above a wall of distorted guitars, tortured cymbals, shimmering reverb, and sizzling synth patches. Needless to say, it is common to push vocals with a high dose of high-frequency hype (through EQ, see Chapter 5). Engineers can get away with this aggressive equalization move except for those instances when the vocal was already bright to begin with: during hard consonants like S and F (and, depending on the vocalist and the language, even Z, X, T, D, K, Ch, Sh, Th, and others). These sounds are naturally rich in high-frequency content. Run them through the equalizer that adds still more high-frequency energy and the vocal zaps the listeners′ ears with pain on every S. It is unacceptable when everyone in the room blinks each time the singer hits an S.
It is a tricky problem. Tracks on the radio have wonderfully bright and detailed vocals. When an audio engineer tries this for the first time in their studio, it comes with a price. While much of the track may be improved, the S′s are too loud, so loud it hurts. Clever compression will solve this problem.
In all of the compressor applications discussed so far, the settings of threshold, ratio, attack, and release have been based on the signal being compressed. What if the compressor adjusted the gain of one signal while ″looking at″ another?
specifically, consider compression of the lead vocal. But instead of compressing it based on the vocal track itself, use a different signal to govern the compression. Called a side chain, a modified vocal signal is fed into this alternative input direct to the level detector (Figure 6.6). The vocal itself is what gets compressed, but the behavior of the compressor — when, how much, how fast, and how long to compress — is governed by the side chain signal. To get rid of S′s, we feed a signal into the side chain that has enhanced S′s. That is, the side chain input is the vocal track equalized so as to bring out the S′s, and deemphasize the rest.
This S-emphasized version of the vocal never makes it into the mix. Listeners do not hear this signal. The compressor does. When the singer sings an S, it goes into the compressor′s level detector loud and clear, breaking the threshold and triggering the compressor into action. The side chain signal is the vocal with a high-frequency boost, maybe a +12 dB high-frequency shelf somewhere around 4–8 kHz, wherever the particularly painful consonant lives for that singer. The side chain vocal can be cut at other frequency ranges; the lows are not needed. The compressor is set with a mid to high ratio, fast attack, and fast release. The threshold is adjusted so that the compressor operates on the loud S′s only, and ignores the side chain vocal the rest of the time.
Figure 6.6 Signal flow of de-esser.
The result is a de-essed (as in, no more letter S) vocal. The compressor instantaneously turns down the great sounding vocal every time it sings an S. In between the S′s, the compressor does not effect the level of the vocal. This vocal can be made edgy and bright without fear: use EQ to boost the highs, knowing the S′s will be reliably pulled down in level at just the right instant and by just the right amount using a de-esser.
It is generally most accurate to think of compression acting on the amplitude envelope of the signal, not the fine, cycle-by-cycle detail of the waveform. Picture the waveform when the digital audio workstation is zoomed out looking at a few bars of music, not zoomed in looking at a few individual cycles of an oscillation. All of the compression effects discussed above happen at the envelope level. But sometimes compression is made to attack and release so quickly that it chases individual peaks and valleys of a waveform (Figure 6.7).
With a threshold set below the peak amplitude of a sine wave, and an attack and release setting faster than a fraction of the period of the sine wave, the compressor will attenuate during each peak (positive or negative) and uncompress in between the peaks. The output from the compressor is clearly no longer a sine wave. If the ratio is high enough (as was used in the creation of Figure 6.7), the sine wave starts to turn into a square wave possessing the same fundamental frequency as the uncompressed sine wave, but containing all the additional harmonics known to exist in a square wave (see ″Complex Waves″ in Chapter 1). In this application, compression is used to generate related harmonics, hopefully creating a pleasing kind of harmonic distortion.
Figure 6.7 Distortion through compression.
Engineers apply a strong dose of compression — to an individual track or the entire mix — for the well-loved effect of distortion. There is something visceral and stimulating about the sound of distortion that makes the music more exciting. The distortion typically dialed in on most electric guitar amps adds an unmistakable, instinctively adrenalizing effect. Compression, with settings that deliberately modify the detailed shape of the waveform, creates a kind of distortion; it seems to communicate an intense, on-the-edge, pushing-the-limits sort of feeling.
A profoundly effective example of this is Tom Lord-Alge′s mix of ″One Headlight″ by the Wallflowers. At each chorus, there is a compelling amount of energy. It feels right. Listen analytically, not emotionally, and one finds that there is no significant change in the arrangement. The drummer does not start swatting every cymbal in sight. No wall of additional distorted guitars comes flying in. Jakob Dylan′s voice is certainly raised, but it is well short of a scream. Mostly the whole mix just gets squashed (compression has many nicknames), big time. A critical listen suggests that parts of the song actually get a little quieter at each chorus, with the compression across the lead vocal and the entire stereo mix pushing hard. But musically, the chorus soars. That is the sort of compression that sells records.
6.5 Advanced Studies: Attack and Release
Attack and release parameters are frequently misunderstood. Mastering the sonic implications of these parameters can help the engineer tap into the full creative potential of compression.
The concept behind attack time is straightforward: It should describe the length time it takes for the compressor to compress — the time it takes the gain change to reach the necessary level so that the user-defined ratio has been reached. An attempt to plot the gain of the compressor as it goes from its fully uncompressed state to fully compressed reveals the problem (Figure 6.8). The compressor might march directly from uncompressed to compressed (Figure 6.8a). It might begin compression slowly and accelerate into full compression (Figure 6.8b). It might begin attenuation quite quickly and approach the ultimate compressed state asymptotically (Figure 6.8c). It might wait some specified period of time and then snap instantly into compression (Figure 6.8d). Or, it might follow any path, however random, from uncompressed to compressed (Figure 6.8e). Attack time defies a single number description. Release time presents the same issue. The subtle variation in the shape of the gain change can have an audible result. In fact, this is part of what distinguishes one make and model of compressor from another.
While no single number fully describes attack and release parameters, many compressors offer a knob or switch for each, labeled ″attack″ and ″release.″ Each manufacturer reduces it to a single, adjustable parameter so that audio engineers may interact with this setting. Any audio student′s understanding of attack and release will depend in part on this set value. As discussed below, it depends on many other variables. The frustrating conclusion is that the attack and release parameter values are not useful on their own. They must be considered in the context of the other compressor settings, the type of gain change element, and the type of signal being compressed. An attack time setting of 4 on one compressor is not equivalent to an attack time setting of 4 on a different make or model.
Drivers can rely on the speedometer to accurately quantify the speed of the automobile, no matter who made the car, or what year it was manufactured. No such standard exists for attack and release values. They are best understood as indicating a relative range, from fast to slow, for the attack and release rates. On an absolute scale, they are essentially meaningless to the practicing sound engineer.
Figure 6.8 Time-based definition of attack. A single number description is incomplete.
6.5.2 VISUALIZING ATTACK AND RELEASE
In order to better understand the all-important attack and release settings, consider the contrived demonstration signal in Figure 6.9. This signal is a burst of pink noise that begins abruptly, sustains briefly, and then decays. It is relatively stable in amplitude during the sustain portion, and the decay is chosen to appear linear. Compression with various attack and release time settings leads to new, and potentially useful, amplitude envelope shapes.
In all of these examples, the threshold is set low so that the sustained portion of this demonstration signal is well above it. The ratio is generally high (6 : 1 or more) so that the compression effect is clearly demonstrated. For the figures that follow, the top plot will always be the original test signal, uncompressed. The middle figure is the signal after compression. The bottom curve shows the gain of the compressor in reaction to the signal, per the attack and release settings discussed.
As a starting point, consider compression with slow attack and slow release settings (Figure 6.10). Because of the relatively slow attack time setting, it takes a visibly obvious amount of time for the compressor to reach its desired amount of attenuation. The resulting compressed signal now has a modified amplitude envelope, one with something of an initial transient in front of the overall signal.
Figure 6.9 Test signal for understanding attack and release.
Figure 6.10 Slow attack and slow release.
The approach of Figure 6.10 is what was used to sharpen the snare envelope shown in Figure 6.4. The compressor has been used to mold a sharp attack onto the signal, which, depending on the exact setting of attack time, might be perceived as a ″tick,″ ″click,″ or ″thump″ at the beginning of the sound. Sounds lacking a noticeable attack can be sharpened so that they cut through a crowded mix. Slow the attack time just enough, and that ever-elusive ″punch″ that rock and roll so often favors — especially on drums and bass — is found. From a high-frequency-oriented ″click″ to a low-frequency-oriented ″punch,″ attack time is the key parameter engineers use to alter the attack portion of the audio waveform.
Speed the release up to a fast setting (Figure 6.11), and notice that the previously observed modification to the initial part of the waveform is gone. Welcome to the clumsy user interface of the compressor. By modifying the release setting (from slow to fast), the audio engineer changes the way the signal responds to the attack setting. The fast release instructs the compressor to uncompress as fast as it can. Therefore, it constantly keeps attempting to restore gain every instant the signal falls back below the threshold. Keep in mind this noise signal is a randomly changing waveform with a distribution of energy across the audible spectrum. This waveform swings back and forth through zero amplitude with a frequency ranging from 20–20,000 cycles per second. The fast release setting drives the compressor back toward unity gain for that portion of the noise signal that is below the threshold.
Figure 6.11 Slow attack and fast release.
The sharpened attack of the waveform relies on a slow release time setting to even exist. The compressor must be slow enough in its release that it stays compressed for the duration of this signal, once the attack time setting has allowed the compressor to reach its specified ratio and achieve full compression.
Consider now compression with fast attack and medium release (Figure 6.12). The attack portion of the amplitude envelope of this test signal after compression is now very much smoothed out. Speeding up the attack time setting can have the effect of removing any rough texture or transient properties of the signal, giving it a smooth, steady, unobtrusive entrance. This approach can be used to dull the attack of a drum, such as the snare drum, for example. Fast attack compression can make an overly angular acoustic guitar that is distracting in the mix sound smooth and more integrated into the mix.
Notice the change in the decay of the test signal, when subjected to this fast attack, medium release compression. The previously straight-line decay has been reshaped. Zooming in on the late portion of the audio waveform (the lower portion of Figure 6.12) reveals two slopes during the decay. The amplitude envelope remains flat during the sustain of the signal. While decaying, the compressor releases (turns the signal back up, when it can), leading to a shallower slope. Finally, once the signal is fully below the threshold, a still steeper decay slope is observed as the signal reaches its end after being compressed and subjected to some make-up gain. Increase release to a faster setting and the decay is further modified (Figure 6.13). A linear slope becomes a complex curve.
Figure 6.12 Fast attack and medium release.
Figure 6.13 Altering amplitude envelope.
The release time parameter provides a way (if the attack time is fast enough) to change the decay of the signal. Human hearing is incredibly sensitive to the reshaped amplitude envelope that compression can create. If a flute and a violin play the same note, we still distinguish the flute from the violin in large part because of the amplitude envelope templates we have in mind for these instruments. Modifying the shape of the onset and the |decay of the envelope allows the engineer to challenge the listener′s basic assumptions about timbre. In addition, the release parameter on the compressor enables the recording engineer to extract and make audible some previously difficult to hear low-level portions of the signal — highlighting important details, exaggerating intimacy, and heightening realism.
6.5.3 PARAMETER INTERDEPENDENCE
The parameters of compression (very much unlike the parameters of a parametric equalizer, Chapter 5) are a rather clumsy set of controls for achieving an automatic reduction in signal amplitude. Engineers must pay careful attention. For example, if the signal is not exceeding the threshold, no adjustment to ratio, attack, or release will have an effect.
Figure 6.14 Slow attack and medium release.
There is a particularly challenging tug of war between the attack time and release time settings. Consider the slow release setting already shown in Figure 6.10. It may be difficult, if not impossible, to hear the result of any small adjustments to release time (compare with Figure 6.14), because the bulk of the release happens after the audio signal has ended. With no signal present, how can one hear a change to the compression effect caused by a modification to the release time? Moreover, if the onset of the next portion of the waveform occurs before the compressor has released sufficiently, any desired reshaping of the onset of the next waveform may not occur or be audible. That is, the effects of any attack time compressor settings may not be revealed because of the current release time settings.
Figure 6.10 demonstrates the sometimes useful sharpening of the onset of a waveform that a slow attack time can create. Figure 6.11 points out that slow attack time does not guaranty this effect will occur. In this case, because the release time is so very fast, the compressor uncompresses fast enough to prevent that sharpening of the waveform. An engineer adjusting compressor attack times in search of the sharp waveform of Figure 6.10 will do so in vain, if the compressor′s release time is as fast as that shown in Figure 6.11.
Attack settings may make release adjustments difficult or even impossible to hear. Release settings can obscure or make sonically invisible any adjustments to the attack parameter. Depending on the type of signal present and the particular parameter settings in use, some adjustments will have no effect, or might have an unintended result.
The reader should know that the author has wasted many a late hour in the studio puzzling over compression. It is a difficult effect to hear in the first place, and the sometimes counterintuitive and surprising interaction of compression parameters sneaks up on all engineers.
Just as earnest engineers new to the field start to get their heads around the twisted logic and sonic surprise of compression generally, and attack and release specifically, yet another layer of complexity appears. Most, if not all, compressors have some amount of program-dependent behavior. That is, even as the most effective attack and release values are set up on a compressor, the compressor then has the audacity to flex those attack and release parameters based on the signal being compressed. Program-dependent compression, while making compression a more challenging effect to master, is a feature that generally leads to more musical results.
It is not unusual for compressors to treat transients with faster attack and release times than steady-state signals, even when the audio engineer has not modified the attack or release settings. On some models, changing the ratio modifies the threshold.
Moreover, the attack and release trajectories are sometimes a function of the amount of gain change currently implemented by the compressor. Slow settings as the compressor initiates compression are often desirable. However, when a compressor is in the midst of compressing a signal, already applying say 6–12 dB of attenuation, the attack and release times might start to increase. More intense attenuation begets quicker reaction. In the meantime, the attack and release parameters themselves, as set by the engineer, remain misleadingly fixed on the faceplate of the device.
The attack and release settings set by the user amount to starting points driving the general behavior of the compressor. They rarely represent rigid values strictly followed by the compressor at all times.
One of the few times compression can be clearly heard, even by untrained listeners, is when it starts pumping. Pumping refers to the audible, unnatural level changes associated primarily with the release of a compressor. The audio signal contains material that changes level in unexpected ways. It might be steady-state noise, the sustained wash of a cymbal, or the long decay of any instrument holding a note for several beats or bars. Listeners expect a certain amplitude envelope: The noise should remain steady in level, the cymbals should decay slowly, etc. Instead, the compressor causes the signal to get noticeably louder. This unnatural increase in amplitude occurs as a compressor turns up gain during release.
Of course compressors always turn up the gain during release. Pumping is the generally unwanted, audible artifact of compressor release revealed by the slow or unchanging amplitude of the audio signal being compressed. The fun part is that the release value is not necessarily too slow or too fast. The problem may be because the release parameter on the compressor is too, um, medium.
A much faster release time could remove pumping by having the compressor release immediately and unnoticeably after the compression-triggering event (as in Figure 6.11). A snare drum hits. The compressor attenuates. The snare sound ends immediately. The compressor releases instantly. If this all happens quickly on each snare hit, the snare sound itself makes it difficult to impossible to hear the quick releasing action of the compressor (see ″Masking″ in Chapter 3). The result is a natural sound, despite compression. The release happens so soon after the snare hit and occurs with such a steep release slope that pumping is not easily heard.
More often, a slower release time is a good option for removing unwanted pumping. Slow the release time substantially, so that the level change of the compressor through release takes a couple of seconds or longer, and the gradual increase in gain by the compressor is slower than the gradual reduction in amplitude during the decay of the sound. The result is a natural-sounding decay of the audio signal, even as the compressor slowly turns up the level. The cymbal takes longer to decay, but it still sounds like a naturally-decaying cymbal. Very slow release is another way to achieve compression without pumping.
Beware of pumping whenever a steady-state, near-steady-state, or sustained sound occurs within the signal you wish to compress: cymbals, long piano notes, tape hiss, electronic or acoustic noise floors, synth pads, whole notes, double whole notes, or longer. Signals that have a predictable, slowly changing amplitude envelope can inadvertently reveal compression, forcing the recordist to abandon compression altogether, back-off the degree of the effect, or at least modify the release settings.
Note, on the other hand, that the pumping artifact may, on occasion, be an interesting effect. If each snare hit causes the cymbals to pump, the result can be an interesting, unnatural envelope in which it sounds as if the cymbals have been reversed in time. Used sparingly, such an effect can add great interest to the production.
6.6 Learning Compression — Some Personal Advice
This single device, the compressor, is used for a wide range of very different sounding, difficult-to-hear effects, using counterintuitive parameters that are a function of the type of signal being compressed, the amount of compression occurring, and the type of components or algorithms within the device. Compression is a difficult effect to master.
6.6.1 UNNATURALLY DIFFICULT TO HEAR
There is nothing built in to human hearing that makes it particularly sensitive to the sonic signatures of compression. Humans are generally pretty good at hearing pitch, easily identifying high versus low pitches in music. Humans react naturally to volume. Without practice, any listener with healthy hearing can separate a loud sound from a sea of quieter sounds. Our hearing mechanism can do deeper analysis, for example, identifying the source of a sound and the location and approximate distance of that source, without conscious thought. The hearing system figures out the angle of arrival of sounds by analyzing, among other things, differences in amplitude and differences in time of arrival at the two ears. Our hearing system tells us which way we need to turn our head in order to face the sound source. Human hearing also extracts seemingly subtle (and darn difficult to measure with a meter or oscilloscope) information from a signal, such as the emotional state or personal identity of a sound source. Who is calling out for us? Are they happy, sad, or angry? Identifying pitch, volume, arrival angle, emotion, and identity are all easier than riding a bicycle; humans do it naturally, instinctively, preconsciously, with no deliberate, intellectual effort or training.
This simply is not the case for compression. There is no important event in nature that requires any assessment of whether or not the amplitude of the signal heard has been slightly manipulated by a device. Identifying the audible traits of compression is a fully learned, intellectual process that audio engineers must master. It is more difficult than learning to ride a bicycle. It is possible that most other people, including even musicians and avid music fans, will never notice, nor need to notice, the sonic fingerprint of compression. Recording engineers have this challenge to themselves.
Early in their careers, most engineers make a few compression mistakes. Overcompressing is a common problem. An effect that is difficult to hear must be overdone to become audible. Compression falls easily into that category of effect.
Young engineers will try compressing until it is clearly audible, and then backing off, so as not to overdo it. Not a bad strategy. At the early stages of learning compression it is not unusual to find that sometimes over-compression is not noticeable until the next day. The effect of compression is at times quite subtle and at other times quite obvious. Spending all day mixing one song, with ears wide open, can make it hard to remain objective. A fresh listen to the mix the next day can be an effective education. Try to learn from these mistakes, but cut yourself some slack. It is not easy.
Wait, it gets worse: It is often the goal of the engineer to set up the compressor so that its affect is inaudible. For many types of compression, we wish the effect to be transparent. Engineers carefully adjust threshold, ratio, attack, and release until they do not even notice it working. Do not let anyone tell you otherwise — tweaking a device until it sounds so good that you can not even hear it isn′t easy.
You are gonna hear people say, ″Hear it, man? It′s beautiful. That′s definitely the Squishmeister G6vx and it sounds so cool.″ And you might think, ″I don′t hear it. What are they talking about? How do they know this? How do they know which compressor it is when I don′t even know it′s compressed? Should I get out of audio and perhaps become a banker?″ Again, some compression is difficult to hear and requires experience. Perhaps they have had the chance to hear this kind of compression before. All you need is time between the speakers immersed in compression of all kinds and you will pick it up. Keep in mind also — and I speak from experience here — that some people in the music business are full of hooey.
Learning to use compressors, and knowing with confidence when to reach for a specific make and model, is a nearly impossible challenge very much like microphone selection.
At first, it makes sense to imitate other engineers whose work we admire. Microphone X on snare drum, microphone Y on female vocals, etc. We take in the vast complexity of the sounds that result over many recording and mixing experiences. Then, when we reach for a different microphone in a familiar application, the sonic contribution of the microphone becomes more obvious.
So too with compressors. We observe successful use of certain compressors for vocals, others for snare drums, and still others for electric bass. As a starting point we can do the same. Once we feel we are getting satisfying sonic results using these specific compressors for these specific applications, we can start to branch out and explore, reaching for different compressors in tried-and-true applications. The sonic fingerprint of the compressor begins to be revealed. Over time, we start to internalize the general sound qualities of every compressor we own, and every make and model of every studio we hire. That is our task as advanced users of compression.
Compression is not a single effect. It is used to prevent overload, overcome noise, increase perceived loudness, improve intelligibility and articulation, smooth a performance, alter amplitude envelope, extract ambience and artifacts, de-ess vocals, and add distortion. In order to correctly learn from the way other engineers use compression, we need to know which effect they sought to achieve. There is no such thing as a vocal compressor or a snare drum compressor. When an engineer reaches for a specific compressor for a specific instrument, they are not demonstrating the universally correct compressor for that instrument. When we learn by watching and listening to others, we need to somehow know, or at least have a good guess, what they were trying to accomplish from a production point of view. It is rarely appropriate for an assistant engineer to interrupt the productive and creative flow of a session and ask this directly. The discreet assistant watches, guesses, experiments, and eventually learns. If you have a good relationship with the seasoned engineer, you might be able to have a discussion on a specific compression strategy in a calm moment after the session. Try to find what they were going for when they patched-up that particular compressor on that particular track at that particular moment. That can be the seed for your eventual success with that kind of effect on that kind of compressor.
Obtaining a complete understanding of compression grows more complex still. Every compressor with adjustable parameters offers a great range of sound qualities. Compressors do not have a single personality for us to discover and codify. In fact, for every make and model of compressor we wish to skillfully wield, we need to develop an understanding of its unique capability across the adjustable parameters. Memorize each compressor for low, medium, and high ratios, across fast, medium, and slow attack times, using fast, medium, and slow release settings. There is a frustratingly large range of permutations to work through. Experienced engineers have done this already. The reader is encouraged to choose a couple of compressors and develop this kind of deep knowledge for themselves. With experience, you will find this task less daunting. Then as you seek to become proficient with additional compressors, you will assess their behaviors and sonic traits relative to the compressors you already use comfortably.
This is not difficult to do, but it is a slow process. There is no substitute for experience. Any pianist can tell within an eighth note whether the piano is a Steinway, Bösendorfer, Baldwin, Falcone, or Yamaha. Any guitarist can identify a Stratocaster, Telecaster, Les Paul, and Gretsch. Seasoned guitarists can also make good guesses as to the pickups used, the gauge of the strings, and if an alternative tuning is being used. Audio engineers need to develop a similar, instinctive ability to identify specific compressors. The distinctions, indescribable at first, become second nature.
6.7 Achieving Compression Success
Dialing-in the right settings on the right compressor as needed in the course of a production is not easy. Experience is essential, but that is about as fair as requiring experience for a job before you can get a job. At first, you are stuck. Assert yourself with confidence even when you are new to compression. Listen carefully to your productions and compare them to recordings you admire. Assist more experienced engineers whenever you get the chance.
The process of using a compressor begins with vision and a strategy: which type of effect to you wish to achieve? Recall again that compressors are capable of a vast range of results; target a specific effect or two from the long list of possibilities: clipping prevention, noise suppression, loudness maximization, intelligibility and articulation enhancement, performance smoothing, amplitude envelope alteration, ambience and artifact exaggeration, de-essing, and distortion.
The type of compression effect selected, choose the make and model that is most effective at achieving this. It takes experience here. As with microphone selection, you will get better at this over time. It feels like a trivia contest at fist. Through experience on many different compressors, though, this step will become easy and intuitive.
Next, set the parameters to settings appropriate for the type of effect. With an understanding of the overall strategy for the effect, threshold, ratio, attack, and release can be preset to values near where they need to be before any audio is run through the compressor. For example, sharpening the attack portion of the amplitude envelope requires a low threshold, high ratio, medium attack, and medium to slow release.
Now, at last, listen carefully to the resulting compression and tweak the parameters until you have fine-tuned the effect to what is needed. Glance at the meters on the compressor to make sure you are not causing un-wanted overload and to confirm the intended amount of gain reduction is occurring.
That is all there is to it. If no amount of parameter adjustment leads to the desired effect, reach for a different compressor. As with microphone selection, the initial choice of equipment determines the range of possible results. Getting this wrong can make certain goals unachievable. Especially early in your career, expect compression to be a highly iterative, sometimes frustratingly slow process. Steady progress is sure to make the overall quality of your productions improve.
Artist: Sheryl Crow
Song: ″Everyday is a Winding Road″
Album: Sheryl Crow
Label: A&M Records
Year: 1996
Notes: The vocal ranges from impossibly intimate to triple forte, yet the objective loudness stays essentially the same. We feel the emotional dynamics, but the vice grip control over level guarantees every syllable is understood.
Artist: Death Cab for Cutie
Song: ″Marching Bands of Manhattan″
Label: Atlantic Records
Year: 2005
Notes: Listen to solitary piano note signaling the end of the tune. Compressed for unusual decay.
Artist: Los Lobos
Song: ″Mas y Mas″
Album: Colossal Head
Label: Warner Bros. Records
Year: 1996
Notes: Heavy compression is part of the sound of the production/engineering team of Mitchell Froom and Tchad Blake. While the entire album offers a wealth of examples, this tune is chosen because, well, the tune grooves hard. Compression is heavy on the vocal; no matter how much he screams, the vocal stays locked-in at one level. It is on the guitars; listen to the spring reverb rise up in level between the wailing notes on the rhythm guitar panned right.
Artist: Liz Phair
Song: ″Polyester Bride″
Album: whitechocolatespaceegg
Label: Matador
Year: 1998
Notes: Snare is compressed for, among other things, exaggerated sustain. First snare hit of the tune is particularly revealing of the effect, before the rest of the mix kicks in.
Artist: The Wallflowers
Song: ″One Headlight″
Album: Bringing Down the Horse
Label: Interscope
Year: 1996
Notes: Lead vocal is compressed aggressively throughout. Note especially chorus vocals with added distortion, brought about in part through extremely fast compression. Snare also offers a classic demonstration of extracting maximum timbral detail, in part, through fast release compression.