TCRM 17 defined and discussed the various forms of dynamics processing and their control parameters. Included in this discussion were compression, gating, limiting, and expansion as well as threshold, ratio and both attack and release times. Though common basic uses were mentioned, specific uses were saved for this column; let’s go through these now.
Strategies, not recipes!
Dynamics processors are capable of fulfilling a wide range of studio chores. They can be used to aid in the tracking and mixing process, address technical issues, shape an instruments sound (timbre), or create some fairly far-out effects. Their use can vary from subtle to glaringly obvious. Due to their great flexibility, they are an essential part of any recording or production studio.
Unfortunately, when used improperly, they often do more damage than good. Many times, this can be traced to the blind use of presets, myths (stories of "magic settings"), and/or forgetting to listen critically and make adjustments. It cannot be overemphasized that finding exact settings must be done on an individual, case-by-case basis.
Major variables that influence dynamics usage include: instruments used, performance, acoustic contexts, personal preferences, musical context, microphones, mic preamps, signal processing, gain structures, and exact make/model of compressor. Any one of these factors is enough to change exactly how a compressor should be set.
But if all of this is true, and specific settings cannot be given, what use is there in discussing how dynamics can be used? Plenty. What can be stated are guidelines regarding parameter ranges for various uses, and the processes/criteria involved in finding the exact settings appropriate to your needs.
If you've been reading this column from the beginning you may remember our mantra: "Recording is an art." Repeat it now, softly to yourself, several hundred times....
The use of dynamics processors is also an art and subject to personal style and tastes. Its use should be an expression of your musical intention and sonic desires.
Compressors in the mix
On of the most common uses of compression is to even out the dynamics of a performance. This can serve two functions: to create musical consistency, and/or to help an instrument stand out in a mix. By setting a compressors threshold to trigger on hotter (louder) notes but leave softer ones alone, the relative level from note to note becomes more even. Since all notes are now closer in level, it is less likely that some will be buried by other instruments while the others stick out above the mix.
This is often an absolute essential for vocal tracks. With medium length attack and release times (20-40 and 60-80 ms respectively), the average levels of the vocals can be evened out and brought up while stressing breath and mouth sounds as well as the start of each line. This gives the vocals a more intimate (close) feel. For this, ratios from 4:1 to 8:1 are most common. A faster attack time can be used to even out a vocal performance, aiding in intelligibility but effecting timbre less. Release times may need to be increased to avoid the fade-in (as the compressor stops reducing the signal) being too obvious - a sound known as compression “pumping” or “breathing”.
Many other instruments are also susceptible to uneven (or too wide-ranging) performance dynamics when recorded. This can be due to instrumental or room acoustics, proximity to the mic, electronics, amplifiers, or (dare we consider...) the performers themselves. For example, electric basses and guitars commonly have problems with consistency from note to note or string to string. This is especially true of less expensive, poorly setup/maintained, or inconsistently strung instruments. (Don't kid yourself... this is the state of a vast majority of the instruments out there).
So here's the basic outline for this method:
- insert the compressor
- choose a compression ratio (2:1 for a little... to 8:1 for a lot)
- adjust the threshold until it triggers on some notes... but not all
- set attack and release times for desired effect
- adjust makeup gain to bring the overall levels back up.
Remember, while there are numbers and meters used in this process, they should not be relied on too much. Use your ear. No matter what the numbers say, if the dynamic change in level sounds too conspicuous to you, decrease the ratio and/or increase the threshold.
Using compression to sculpt timbre
Timbre is the particular sonic character that’s individual to an instrument or performer. The perception of timbre depends on a number of factors, the most important being spectra/time, and dynamic contour, called envelope. This envelope is often broken down into four main time/level elements. These are (in chronological order) attack, decay, sustain, and release (ADSR).
Here, attack and release are similar ideas to the attack and release time settings on a compressor. They are the begging and end of a sound, and the beginning and end of the dynamic effect. In fact, in some instances, processor settings can be made to affect the envelope of a sound, thereby changing its very character. Below I’ll outline how this might be done. Before starting, however, it should be noted that the exact nature of the timbral shift also depends highly on the sonic influence of the exact design and implementation of the processor.
When careful attention is paid to finding the right release time, the tail of notes can be sculpted to change timbre or extend sustain. This can do wonders on kick drum or tom sounds when a slightly longer sound is required. When a vocalist is having trouble supporting longer held notes, these can be saved using longer release times. The technique is accomplished by carefully setting the threshold below the strongest portion of the note, but above the fading tail. As the note begins to die away, the compressor slowly begins to stop reducing the level... fading in as the note fades out.
These methods, when used to reshape the end of notes or phrases, are even more effective on previously compressed, sampled or synthesized sounds. This is due to the fact that these sounds are already more consistent in regards to levels and dynamic contour. Because of this, the effects of these highly specific compressor settings will also be more consistent.
While on the topic of gain and release, it should be noted that, since compressors release at the end of notes or musical phases, that the sound of the room acoustics will also be brought up. Later reflections and reverb may be accentuated, as well as other ambient noises.
Of course, attack time settings can also have a tremendous influence on timbre. They can be used to either accentuate or reduce the sonic influence of transients (very fast changes in amplitude at the start of a sound), or to further modify the onset of the body of an individual sound. On many percussion instruments, where the transients are quite strong, using an extremely fast attack (under 3 ms) would reduce transient energy as well as the rest of the sound. By adjusting the attack time to equal the length of the transient portion (usually a 6-14 ms setting), transients are left alone, and the note is reduced. The result is an accentuation of the initial "smack" or "thud" ("whack?"). On guitars, this technique can be used to accentuate the attack of the pick, if desired (attack times from 14-50 ms).
When attack times are longer than any transients, or transient influence is weak, you can achieve further sculpting of the attack, decay, or even sustain of a note. As this is very application-specific, you’ll just have to play around with this one yourself. Before you begin, be aware that if the attack is too long, and the release point is already triggered before the attack gets going, there may be little noticeable effect at all.
Another trick I’ll mention here, because it’s a timbral effect, is particularly suited to work on DAWs or other digital mediums. It’s the idea of inserting two or more versions of compression on the same track. This method can be used to emulate older analog recording and mixing technologies and methods. Many engineers compressed the signal while tracking to help maintain the best S/N ratios. Then, the analog tape medium compressed it further. On mixdown, tracks were often compressed again to aid in balancing. In most digital studios nowadays, these first two compressions are simply not needed, and have been mostly forgotten. To recapture that older sound, try adding them back in - even at mixdown after everything is already tracked.
Expansion and Gating
As outlined in TCRM 17, expansion is really the opposite of compression in that it actually increases the dynamic range of a recording. Instead of reducing hotter signals, expansion serves to increase them (relative to the lower ones). By setting a threshold below all of the program material, but above any background or quantization noise, an expander can be used to decrease noise during times when the performer is not playing. Similarly, and also in opposition to compressors, an expander can be used to reduce the sense of acoustic space between notes or phrases.
On a timbral note, expanders are handy for de-accentuating transients. If the attack time is set to approximately the length of the transient material, then the sound will at a normal level for that moment, but increased for the remainder of the sounds envelope. In this manner you can dial up less beater on a kick, less pick on guitar or bass, less “snap” on the snare, or less “chiff” on winds.
In some ways similar to an expander, the most common use of a gate is as an automatic noise-reducing device. Instead of reducing the lower level signals, however, gates remove them.
Especially noisy tracks, with excessive hiss or hum, quickly draw attention away from an otherwise great performance. Common culprits include guitar amps, unbalanced equipment, and/or faulty (or just ill-advised) levels, cabling and connectors. A gate, however, can be set to stop the noise from passing through when a musical part is tacet (not playing). Do this by setting the threshold a little below all wanted sound, but above the unwanted.
Beyond being simple noise-reducers, gates are also great for tweaking timbre and creating cool effects. They can modify the acoustic context of a performance (by removing reverb tails and room tone), and/or change the actual envelope of a sound. This is especially true on shorter, staccato sounds. As gates have both attack and release settings, these methods are similar to the compression and gating ones described above, but more drastic. Fast release times serve to make very concise snare hits.
Another popular sound is that of the gated reverb, made popular by engineer Hugh Padgham on Phil Collins’ drums in 1980 – check out “Intruder” on the third Peter Gabriel album, and on “In the Air Tonight” on Face Value. Here, a very obvious reverb is inserted on an instrument’s track. After the reverb, the channel is run though a gate whose threshold is set just below the level of the instrument. When the instrument is playing, it has a strong sense of acoustic space. When the note begins to die away, that context suddenly vanishes. It’s a very cool, but definitely “unnatural” sound.
Limiters serve three main functions, as a level safety net, as a drastic level maximizer, and (here I go again?) as a timbral modifier. The first function (safety net) is particularly important in digital recording and mixing. Everyone strives to capture and reproduce good strong signals. But, if these signals go above 0 dBFS, however briefly, a nasty form of distortion occurs, called clipping. Limiters set between -4 to –1.5 dBfs are commonly used to ensure this does not happen.
The second main use, to maximize levels, is most commonly saved for the final mixdown as a form of mastering. By using a limiter to place a ceiling (say –1.5 dBfs) on a signal’s level and increasing the input gain, the average level of the signal is quickly increased. Though the peak level doesn’t increase, the perceived loudness sure will. On the other hand, dynamic range and musical expression are decreased, along with the S/N ratio. If the limiter is pushed too hard, distortion and other artifacts will also become very noticeable. So be very careful not to overdo this form of “mastering” - it is very easy to abuse and will quickly destroy your hard work tracking and mixing.
To make mastering compression/limiting at once both easier and more frequency specific, mulitiband compressors can be used. These dynamics processors react differently depending on which frequency areas have more energy. Again, while these are very powerful, they can also be abused fairly easily and actually reduce your audio quality.
As mentioned already in earlier installments of this column, if you have the budget to hire a professional mastering engineer for this important job, that is usually the better way to go. They have the dedicated equipment, experience, and necessary objectivity to do the job right.
Of course, where there’s a rule, there’s an exception; while extreme compression or limiting parameters can create distortion, that can sometimes be a very cool thing. For instance, some distortion can add interest to a vocal track sometimes and a fast-acting compressor can be used to create a very compelling sound. The key here is to use dynamic processing to create distortion, or any timbral change, intentionally and when there’s artistic reason.
As discussed in TCRM 17, the sidechain (aka. key, trigger, or control) is a secondary input to a dynamics processor that, when active, determines when and how the primary (program) input will be effected. When the threshold is exceeded by the sidechain signal, the program audio is compressed.
De-essing is one of the most popular, sidechain-based compression methods. It is used to diminish the harsh vocal "s" "sh" "ch" and "t" sounds collectively known as sibilance. De-essing is a form of frequency-dependant compression. Here, the vocal channel is split, with one half being inserted into the program input of a compressor, and the other half to an equalizer before being inserted into the sidechain input of the same compressor. The eq is used to find and accentuate the offending consonant sounds. The threshold is then set to trigger compression when those sibilants are present, but not during unrelated consonants or vowels. With fast attack and release settings, the compression ratio can be used to determine the amount the sibilants will be reduced. This basic method can also be used to reduce vocal plosives or low frequency rumbles caused by the breath moving the mic diaphragm. Sounds such as "p" "b" or "f" are common culprits.
There are many dedicated hardware and software deessers available that make this rather complicated signal path and process invisible to the user. Often, the eq curve is determined simply by selecting the singer’s gender (since a woman’s characteristic sibilant frequency will be a bit higher than a man’s). While these work fine in a lot of cases, they cannot be relied upon to work on all singers or singer/mic combinations.
Manually-configured de-essing setups, though more complicated to use, can be much more flexible and need specific. In fact, they can be used effectively on more than just vocals. For example, it is not uncommon for instrumental recordings to exhibit frequency-specific issues which cause certain notes to stick out above the rest. Sidechain "deessing" methods can be used to tame these as well, even if the removed offenders are not “esses”.
Ducking is the name for a method where one signal is used to trigger dynamics processing on a different signal. You may have heard it on live sports broadcasts when the audience noise “ducks” under the commentator’s voice and swells up again during the (rare…) moments when the commentator is silent.
Another common musical example of this is to duck (compress) the bass with the kick drum. Since the kick and bass occupy much of the same frequency range and are often played simultaneously, the two sounds tend to obscure each other and actually reduce the clarity of the beat, as well as the timbres of both instruments. Ducking is a great way to combat this problem; here's how:
- insert a compressor on the bass channel
- send the kick to the sidechain of that compressor
- set threshold to trigger with each kick hit
- set short attack and short to medium release times
- adjust ratio to achieve desired effect without being too drastic
With appropriate settings, the bass part will be reduced momentarily each time the kick is played. This makes room for the kick to come out more strongly and clarify the beat. At the same time, the underlying support and groove of the bass line is maintained, as the average level is unchanged.
Since guitars and vocals share similar ranges and can also compete with each other in the mix, ducking can be used to address these problems as well. Start by submixing the guitars and inserting a compressor on the submix. The sidechain input will be the vocals. In this manner, the guitar levels can automatically be reduced whenever the vocals come in. In some cases, this approach can be taken a step further; the entire music bed is ducked by the vocals, ensuring greater intelligibility and clarity. When vocals are not present, however, the sonic energy of the music is maintained. This method is extremely common in radio, television, and video/film.
Triggered layering or doubling
Some gates are also equipped with sidechain inputs which can be used to create all kinds of cool effects. The most basic of these is triggered layering, where one sound keys the addition of another. This effect is most popular on (though certainly not limited to) drums, especially kick and snare. Snare sounds are often set up to trigger an alternative sampled snare sound or noise bursts to enhance the existing snare sound.
On a kick, a low-frequency sine tone (or other simple waveform) is sent to the program input of a gate. The sidechain input is then fed the kick. Whenever the kick hits... the gate opens and the low frequency sine tone is let through. This can be mixed with the original kick sound, or used to replace it. Further cool effects can be created by delaying the gated waveform by a meaningful rhythmic duration (like an eighth or sixteenth note). I suppose choosing to delay the dry kick sound could also generate a pretty funky groove....
Another use of this goes as follows: Say you want to trigger a short hit from a synth chord every time the already recorded rhythm guitar plays a chord. Pick the sustaining synth sound and hold down a chord (for as long as the same chord/change applies). Then put a gate on the synth track and set up the rhythm guitar to be the trigger, letting the sidechain open the gate, triggering the desired hits in sync with the already recorded rhythm guitar.
Sidechain techniques are sometimes also required to create a cleaner version of the “gated reverb” effect mentioned previously. The simple method of gating a reverb return may create very inconsistent results from note to note. Some will have a longer reverb tail, while others will be chopped very short. This is because instruments, performers, reverb algorithms, recording technology, and acoustics cause variances in both dynamic and frequency content. The threshold of the gate is not broken at the same points in the envelope of the sound each time it occurs. To avoid this problem, the dry signal can be used to trigger the gate on a reverb-heavy version of the same material. Even greater consistency will be achieved if the dry signal is compressed just before being inserted into the sidechain. The more even the level of the triggering sound is, the better the effect will be.
Many of the functions of dynamics processors are now also possible using the volume (fader) automation common to DAWs. Drawing in, or performing, all of the necessary changes to sculpt each note and nuance, however, would be time consuming (to say the least). Though these methods may be tempting to those who have learned their recording and mixing skills almost exclusively on DAWs, learning proper use of dynamics processors is truly invaluable. The time they can save, and added creativity they can inspire will make any time spent experimenting and exploring their use pay back tenfold (probably much more).
I hope these concepts give you some new ideas and get you experimenting with dynamic effects. Next time I’ll dive into common effects such as reverb, flange, phase, chorus, delay, emulators, and others….
John Shirley is a recording engineer, composer, programmer and producer. He’s also a Professor in the Sound Recording Technology program at the University of Massachusetts Lowell and chairman of their music department. You can check out his wacky electronic music CD, Sonic Ninjutsu, at http://cycling74.com/products/c74music/
Supplemental Media Examples
The following audio examples are meant to demonstrate dynamics processing techniques as discussed in this article.
Dynamics for Mixing Basics
The first thirteen examples here demonstrate techniques for using compression to clean up the mix and make each part more distinct and clear. First, let’s look at the all-important vocal line.
Original vocal line without any compression. TCRM18_1.wav
Vocals with a fast attack (less than 1 millisecond) and medium release (45 milliseconds) compression to even out the phrasing and make each word more equal in volume. The ratio here is around 5:1 and the threshold is set so that quieter moments are not effected at all but the louder words are compressed. TCRM18_2.wav
Now I’ve added another compressor to accentuate the attack of each word and the breath and mouth sounds in preparation to sing as well as at the end of words and phrases. This makes for a closer sounding recording. Here, the attack is 18.5 ms and the release is 80 ms. The ratio is 4.5:1. TCRM18_3.wav
Finally, I’ve added a deesser to tame some of those “s” and “t” consonants. TCRM18_4.wav
Now let’s look at compression for the bass. First, the bass part without any compression. TCRM18_5.wav
Now the bass is compressed so that the performance dynamics are more even note-to-note. This is done by using a fast attack compression with medium release similar to the approach used for vocals in TCRM18_2.wav. TCRM18_6.wav
Finally, the bass level is momentarily reduced each time the kick hits. This makes the accent of the kick come out more clearly in the mix and is accomplished by inserting a compressor on the bass track but using the sidechain (key) input, which is bussed from the kick track, to trigger the compression. TCRM18_8.wav
Now let’s look at how compression can effect the sound of the snare on this track. First, let’s listen to the snare without compression. TCRM18_9.wav
Now the snare hits are evened out. TCRM18_10.wav
Finally, the attack of the snare is accented and the resonance held slightly longer. TCRM18_11.wav
Finally, let’s compare a mix of this song with and without dynamics processors. First, without. TCRM18_12.wav
Then with all of the basic compression and ducking put back in (as in TCRM18_2.wav through TCRM18_11.wav). TCRM18_13.wav
Dynamics Processing for Timbre
Now, let’s look at more aggressive uses for dynamics that change instrumental timbre more noticeably. Examples 14 through 27 take this approach.
Using a somewhat different mix of the kit, now we’ll see other ways the snare could be treated with dynamic processors. First, dry. TCRM18_14.wav
A gentler, and slightly less distorted, snare sound comes from using a soft knee. TCRM18_16.wav
Now, the dry snare sound is run through a very large reverb. TCRM18_18.wav
Next, a gate is inserted after the reverb. Oh, the 80’s were so cool! TCRM18_19.wav
By changing the gate settings, a different timbre can be sculpted. Here, a longer release is used. TCRM18_20.wav
A gate could also be used on the dry snare to create another cool sound. TCRM18_21.wav
Extremely fast attack and release times, and/or extreme ratio versus threshold settings, can actually cause waveshaping and distortion. TCRM18_22.wav
Next, a solo snare sample is repeated, first dry and then using the sidechain on a gate to trigger a noise burst along with it. Dry: TCRM18_23.wav
With gated noise (does this effect sound familiar?) TCRM18_24.wav
This same gate/sidechain idea has also been used to great effect on the kick by triggering low sine tones when the kick hits. Here’s the dry kick. TCRM18_25.wav
Now with a 49 Hz sine tone. TCRM18_26.wav
Lastly, with both 49 and 24 Hz tones. TCRM18_27.wav
Finally, a sample of the vocal line again, but this time with the limiting/waveshaping (fast attack and release with high ratio). TCRM18_28.wav
Addressing Technical Problems
Here, a guitar from a different song demonstrates the use of gates for noise reduction in a track. The first example is without a gate. TCRM18_29.wav
Now the same region with a gate and extra reverb to make up for some truncated reverb tails. TCRM18_30.wav
Notes on Threshold settings
Though approximate attack times, release times, and ratios have been given above for various approaches to dynamics, little specifics are given regarding threshold numbers. This is because the setting of an appropriate threshold is not only determined by the type of function you are trying to accomplish, but the relative levels of the incoming program or sidechain audio. Because of this, settings cannot be generalized, but must be set as appropriate to each individual situation. Below are some examples of how this is done and what happens if threshold settings are set haphazardly.
The settings used to make TCRM18_2.wav were intended to even out the performance dynamics of the vocals without changing timbre too much or sacrificing clarity. To do this, the threshold should be set such that compression is triggered only on the quieter words or phrases and not on the louder ones. Here’s that setting: TCRM18_31.wav
But if the threshold is set too far below this, it triggers on almost everything, so the effect is not selective between loud and soft passages. Here, the threshold is reduced to 20dB below the first example. Notice how how timbre is changed and the obvious, odd changes in dynamics. TCRM18_32.wav
The snare example of TCRM18_11.wav was to demonstrate how the attack of the snare can be accentuated using compression, and the sound slightly elongated. If the settings are right, the basic timbre of the snare recording is still fairly natural sounding. For this, the threshold should be set so that the effect happens on each snare hit (it helps if the level-evening compression discussed above comes beforehand), but gain reduction is not too severe (generally from 3-8 dB max) Here’s another example of that: TCRM18_33.wav
But if the threshold is brought too low, the result can drastically alter the timbre of the sound, creating a very artificial snare sound. Here, note how the attack is over-exaggerated and the swell of the release sounds synthesized. (This can be cool if it is your intention) Again, this example uses a 20dB lower threshold than TCRM18_33.wav. TCRM18_34.wav
Using the same approach to the kick to accentuate the attack can add clarity to the beat and draw attention to the kick. TCRM18_35.wav
But again, if the threshold is set too low that attack accentuation can be extreme. Here, the attack is certainly accented, but the body and sense of low frequency is lost. (Again, this example uses a 20dB lower threshold than TCRM18_35.wav.) TCRM18_36.wav
Special thanks to The Bay State for the use of the raw tracks to this demo recording of “Can’t Stop this Now” for audio examples 1 through 28. All material used by permission; all copyrights reserved. Demo recorded by Michael Testa.
The Bay State is Tom Tash, Drew Hooke, Susanne Gerry and Evan James. Check out their music on iTunes (including the commercial release of “Liars” or visit them on facebook at: http://www.facebook.com/thebaystate
Special thanks also to recording engineer Bernie Mack and the band Deuce for the use of the guitar track from the demo for “Baby Baby” which was used for examples 29 and 30.