from an earlier post
Understanding Compressors and Compression
from an article by Barry Rudolph
A compressor consists of four basic parameters:
Ratio
Ratio is a way to express the degree to which the compressor is reducing dynamic range. Ratio indicates the difference between the signal increase coming into the compressor and the increase at the output level. A ratio of 10:1 would mean that it would take an increase of 10 dB coming into the compressor to cause the output to only increase 1 dB. Ratio is a constant value, as it doesn't matter how much compression is taking place; the ratio of the input change to output change is always the same. Compressors and limiters are really separated only by a loose definition: Generally, compressors have compression ratios up to 8:1, while limiters have ratios higher than 8:1. Most professional compressors have either fixed selectable ratios (such as UREI's 1176 with presets of 4:1, 8:1, 12:1 and 20:1) or continuous variable
ratios (such as the dbx line of professional compressors).
Some recent compressor designs instantaneously change ratio depending on the program's dynamic content and the constraints of the front panel control
settings.
Threshold
Threshold is the level of the incoming signal at which the compressor amplifier changes from a unity gain amplifier (like a straight piece of wire, theoretically) into a compressor reducing gain. The compressor has no effect on the signal below the threshold level setting. Once threshold is reached, the compressor starts reducing gain according to the amount the signal exceeds threshold and according to the ratio control setting. Threshold level could be thought of as the "sensitivity" of the compressor and is expressed as a specific level in dB. The exact moment the compressor starts gain reduction is called the "knee."
"Hard knee" compression describes this moment as sudden and certain. "Soft knee" or smooth knee compression is a less obtrusive change from simple
amplifier to compressor. Soft knee widens or broadens the range of threshold values necessary for the onset of compression. On quality compressors you can switch between hard and soft knee compression. The amount of gain reduction is measured and read on a standard VU meter whose needle rests on the 0 VU mark. The needle will deflect negatively downward to indicate how much gain reduction is occurring in dB. VU meters are RMS or average level responding and do not indicate fast or peak gain changes. LEDs are also used for VU meters, and they will better indicate peak levels. A well-designed
compressor will have a good meter that reads input level, output level, gain reduction and any excessive peak output with an LED clip indicator. Once the amount of gain reduction is determined, the recording or operating
level is readjusted with the output or make-up gain control on the compressor.
Attack
Attack time refers to the time it takes the compressor to start compressing after threshold has been reached. Typical attack times range from less than
1 millisecond at the fastest to more than 100 milliseconds at the slowest. Attack time settings affect the sound quality in terms of overall perceived brightness or high-frequency content. If you use very fast attack time settings, the compressor will activate very quickly, reducing gain instantly at the waveform level of the sound. Since transient information at the front or attack portion conveys brightness character, especially with percussive sounds, immediately reducing it with the compressor will dull the sound. Selecting a slower attack time will allow the transient portion of the sound to pass through before the compressor starts clamping. However, if the attack time is too slow, ineffective and tardy compressor action may result.
If you compress a snare drum track with a fast attack, you may notice a diminished or shortened attack of the front portion of the snare drum sound. Instead of a good "hit" at the beginning, you will hear a very short "blip." If
the attack time were even faster, you wouldn't hear any snare attack at all. Adjusting the attack control to a slightly slower attack time will lengthen this blip back to the original snare attack length. Engineers use a compressor to get more attack out of a snare drum by using a low threshold and a high ratio. After making up gain with the output level control of the compressor, the attack portion ends up greatly amplified just before the compressor starts squashing the trailing portion of the snare drum sound. I like to mix this "twacked" sound with the original un-processed snare drum track.
Release
Release time is the time the compressor uses to return to unity gain after the input signal has fallen below threshold. The compressor is said to "release" from gain reduction. Typical release times on popular compressors go from as fast as 20 milliseconds to over 5 seconds. Most engineers envision their compressors doing their job of gain reduction quickly
and then releasing quickly to get out of the way. For the most part this holds up for pop recordings, but super fast release times, along with a fast attack time setting, will distort low-frequency sounds, as the compressor is capable of gain change within the period of the sound's waveform. You can demonstrate this by using a very fast release on a bass guitar compressor and have the player play loud and sustained notes. Over-long release time settings are another form of distortion, since gain reduction is "stuck" clamping the sound down for an unnaturally long time. "Pumping" and "breathing" are engineer jargon words for obvious compressor artifacts or side effects with maximum compression. Sudden and usually unwanted
deep gain reduction is called pumping, while a slower return (release) to operating level with a noticeable rise of the noise floor is called breathing. Newer compressor designs have clever predictive and adaptive schemes that reduce these side effects, making the compressor's action nearly undetectable or at least tolerable in most intense gain reduction situations.
Using Compressors in music
Trying to arrive at proper recording level can be like chasing a moving target, especially if you are working with inexperienced musicians or singers. It is easy to just "pack" the signal from a widely dynamic singer into a compressor, crank it way up and call it a day! On the other hand, the creative and caring engineer may use an equalizer in the sidechain of a compressor to selectively compress certain problem frequencies of the singer's voice. The sidechain input is a direct path to the compressor's detector circuitry, where an external signal can also trigger compressor action. Using an equalizer in the sidechain makes the compressor more (or less) sensitive to sounds within the EQ's frequency passband. Other
sidechain compressor applications are de-essing and ducking. De-essing is accomplished by connecting an equalizer to the sidechain of a peak limiter and boosting, with a medium to high Q, the "s" frequencies anywhere from 1.5 to 6 kHz (depending on the singer). Strong "s" sounds are quickly reduced
without (one hopes) too much consequence to the rest of the vocal sound's signal. Ducking is gain reduction that is triggered from a different audio source altogether. The most common usage is to "duck" or lower a music bed (track) whenever a narrative voice-over is active.
Bass guitar
Generally, bass guitars seem to sound better when using some compression. I sometimes rely on the compressor to pull up some additional bottom end and warmth. I like to use a leveling amplifier for bass guitar such as an LA-2, Anthony DeMaria Labs or
a Tube-Tech CL-1B compressor set up as a leveling amp. Based on the song's tempo and/or the bass player's style of playing, I might start with ratios of 5:1 with a medium threshold setting, medium attack and a slower release time. Unless you are looking for a pumpy effect or have a very wild bass player that the producer would like to squash, I rarely compress bass guitar more than about 5 dB max VU.
Vocals
A lot of anguish and pain surround the recording of vocals. Maybe some singer
insecurities, producer apprehensions and overall great expectations weigh on the arrival of a stellar vocal sound. It's about preferences, and vocal sound is subjective and contextual within the backing track. It does not exist on its own unless you are recording an a cappella performance. What I might think is a great vocal recording on a certain CD may be not be what the singer and producer are looking for to convey the emotional import of the artist, the song's lyric or the vibe of the song. At the first vocal recording session, you will want a starting point vocal sound that fits the singer, the song and the production. This starting point sound will work well to capture the immediacy of the moment and hold up later in mixdown. After mic selection, mic preamp selection and EQ setting, a compressor type is decided. Although there is much interaction between all these components, my starting point settings are predicated upon a +4dB level coming from this mic/preamp/EQ chain. For vocals, I tend to use compressors with adjustable attack and release time controls. I like tube-based compressors, although for more difficult control problems I would go to a VCA-based unit. If you were to set the threshold at
about 0 dB, the ratio at 4:1, the attack and release at middle positions and adjust the output level for a good recording level, you'd have about 3 to 6 dB of compression and probably make most people happy. This is just a good starting point, and you should get in there and change those settings to your own taste.
Stereo mix compression
A popular place to apply compression is on the entire stereo mix as it is going to the master tape. Stereo compression is also applied "after the fact," such as in mastering, but there is a big difference. If you are mixing your record into a compressor ahead of the master tape recorder (or DAW or whatever), then you are mixing many sources (tracks or elements), each with its individual dynamic content. Making mix moves (especially big gain
changes) directly affects the compressor's action, which, of course, affects the relationship of all the other mix elements, that is, the whole mix! If you do not use a stereo compressor at mix and compress at mastering, then you have just the dynamics of a 2-channel stereo mix for the compressor to react to. Many engineers and producers mix into a compressor and then have
the mastering engineer compress again, while others prefer to wait until mastering to compress digitally. A good starting place for the stereo mix compressor is a lower ratio with medium attack and release time settings. Set the threshold (and output level) so that you can hear an increase in level (average loudness) when you A/B the compressor in and out of circuit.
Remember, this is just a starting point! You may want to use a higher ratio to control peaks better and/or a lower threshold for more compression and denser overall finished sound.
