This is pasted from an article by dbx...

-bg-

www.thelittlecanadaheadphoneband.ca www.lchb.ca


Compression 101

Compression often performs subtle changes on a signal and can be difficult
to learn. Most types of signal processors, such as reverbs, equalizers, and
delays, are designed to make an obvious change in the sound. But a
compressor's action is much more subtle; when used properly, most listeners
won't be aware that signal processing is being used. Only if you hear the
original dynamic range of a signal and compare it to the compressed version
will the effect be noticeable. yet, compressors are essential in modern
audio work .Almost every lead vocal on a pop record is compressed during
tracking or mixdown. Often the entire stereo mix maybe compressed or limited
during the mastering process; when you hear it on your favorite radio
station, it passes through another compressor before it's transmitted.

A compressor/limiter is essentially an automatic volume control. Imagine an
engineer with his hand on a fader and his eyes on an input level meter .As
long as the meter stays below a certain point (the threshold), he leaves the
fader all the way up and the gain is unchanged. But the instant the sound
gets louder, the engineer pulls down the fader by a certain amount. After
the sound gets soft again, the engineer will push the fader back up. That's
what the compressor is doing, except much faster and more accurately than
humanly possible. Paradoxically, by cutting the peak levels, a compressor
allows you to raise the average level of a sound using the Output control to
make it sound louder. By using the threshold and ratio controls, you can set
a stable sound that will hold its position in the mix, if the singer is
whispering or screaming.

What the controls do - Let's go back to the "engineer with his hand on a
fader and eyes on the meter" analogy. The front panel controls simply tell
the "engineer" what rules he should follow. [THRESHOLD] tells him how high
the input meter can rise before he has to start pulling down the fader: if
it's turned full clockwise, he won't pull down his fader until the highest
red LED comes on; if it's turned counter-clockwise, he'll have his hand on
the fader even before the lowest green LED lights. [RATIO] tells him how far
he should "pull the fader down "when the signal is above the threshold
level: should he pull it down just a little bit (compression) or pull the
fader as far down as necessary to make sure the output level is never higher
than the threshold (limiting)? The [Hardknee OverEasy®] switch affects how
he reacts as signal approaches and travels through the threshold: does he
reduce it exactly by the ratio only after it crosses the threshold, or does
he gradually ease into the full ratio as it passes through? The LEDs of the
gain reduction meter tell you how much the "engineer" is pulling down the
"fader" at any time. If these LEDs aren't on, his hands are in his pockets.

The [ATTACK] and [RELEASE] controls involve the speed of the engineer's
response. Short attack times order the engineer to get his hands on the
fader 1/10,000th of a second after he sees a too-loud signal; long attack
times tell him to let transients less than 1/5th of a second pass. [RELEASE]
tells the engineer how quickly he should push the fader back up again after
a loud signal has stopped; when it's turned counter-clockwise, he pushes the
fader back up instantly, and when it's full clockwise, he'll take longer to
push his fader back up to unity gain.

The [OUTPUT] control is simply a gain control located after our "automatic
engineer in the box."

The [INPUT/OUTPUT] switch allows you to see the levels before the engineer
does his job, or after.

The most important controls are the [THRESHOLD] and [RATIO] knobs. They both
interact to get the effect you want, and that requires some experimenting.
For example, if your average input signal is 0 dB, a ratio of 2:1 with a
threshold of -12 dB will give you 6 dB of gain reduction, as will a ratio of
infinity with a threshold of -6 dB; the latter setting will sound more
"squeezed" than the former.

Avoid common compressor mistakes - Extreme settings will lead to extreme
results. If you set an infinite ratio and turn the threshold down to -40 dB,
the compressor will do what it's being told to do: turn the level way down.
If you then try to compensate by cranking the [OUTPUT] control to its
maximum, you'll amplify the noise of your mixer, EQ, Mic preamp, and the
compressor itself. The noise will fade itself in whenever the input signal
stops, resulting in the classic "pumping" and "breathing" problems. Noise is
present in every system, improper use of any compressor will amplify it to
an obnoxious level.

If the ratio is set to 1:1, it doesn't matter where the [THRESHOLD] control
is: the compressor is being told not to change the gain at all, even if it's
above the threshold level. None of the REDUCTION LEDs will light, and you
may as well have the compressor in BYPASS mode.

Similarly, if the ratio is infinite and the threshold is high, or the input
trim of the mixer or microphone preamp is too low, you will get no
compression (and, if you raise the [OUTPUT] level control, you'll be
amplifying the noise floor).

For low noise operation, make sure your mixer, compressor, and amplifier
settings are set properly .As a general rule, you want as much gain as
possible in the front of the system (at the microphone preamp), so that a
good line-level signal is travelling through the whole signal path. if you
have a weak signal to start with, and then amplify it at the end of the
signal path (by turning the main outputs of the mixer all the way up, for
example) it will be noisy.

When using a compressor on a live P.A. system, improper settings can cause
feedback. Make sure that a channel is well below the feedback point when
there is no gain reduction active. If you hear feedback every time the music
stops, you must lower the overall level of the system.

About stereo compression
For great technical information on stereo processing, see the dbx white
paper called True RMS Power Summing®, written by Roger Johnson, Director of
dbx Engineering. Otherwise read on...

Terms

Threshold - The THRESHOLD knob sets the level where compression will begin.
As long as the input signal level is below the Threshold level, the
compressor will do nothing to the signal. Once the input signal crosses the
Threshold, the compressor will begin compressing at a ratio set by the ratio
control.

Ratio - The RATIO knob controls the amount of compression, which will happen
once the input signal crosses the Threshold level, described above. Ratio
controls how much the input signal will be reduced as a ratio of the input
signal level. For example, if the compression ratio is set for 6:1, the
input signal will have to cross the threshold by 6 dB for the output level
to increase by 1 dB. the maximum setting is typically labeled: 1 (infinity
to 1), and is also called Limiting. This means that the output signal won't
increase at all, no matter how far above the threshold the input signal
goes.

Attack - The ATTACK knob controls the amount of time before compression
starts after the threshold is exceeded by the input signal .The range of
this control can go anywhere from very slow (about 1dB/Sec) to very fast
(about 400dB/Sec). Long attacks are useful for percussive sounds, where
shorter attacks are good for melodic parts like vocals and strings. The
Attack control is also useful for keeping the transients on percussive drum
or bass sounds. Experiment with different short attack times on snare drums
to get more or less of the "stick" attach.

Release - The RELEASE knob controls the amount of time the compressor takes
to stop compressing after the signal crosses below the threshold. the range
of this control can be from approximately 4000dB/Sec to 10dB/Sec. Short
release times are good for percussive, punchy sounds, where longer release
times can make compression less obvious on vocals. Adjusting the release
time may be necessary when using extreme compression and "pumping" or
"breathing" is audible, or if lower level signals after peaks are getting
lost. See the section on pumping and breathing explained below.

Output - The OUTPUT knob controls the level of the compressor's output. The
Output control is useful for making up gain, which was reduced by the
compression circuit, or matching the input level of a mixer or recorder. If
the Gain Reduction meter shows that the input signal is being attenuated
by -6dB, then the Output control generally should be set around +6dB. This
control is disabled if the [BYPASS] button is pressed.

Hard knee/OverEasy® - The OverEasy® switch is used to switch between Hard
and Soft knee compression styles. When the compressor is set for Hardknee,
the compression ratio applies only to signals above the threshold level. If
the compressor is set for Soft knee, the compression ratio gradually
increases from 1:1 to the current selected ratio over a range through the
threshold area, so that the transition from uncompressed to compressed is
more gradual.

The difference between Hard Knee and OverEasy® is more obvious at high
compression ratios. Once the input signal crosses the Threshold, the unit
will compress the signal at the full ratio level. OverEasy® compression is
useful when performing high-ratio compression or limiting on a signal. When
the compression gradually fades in, it doesn't sound as obtrusive as when it
suddenly starts limiting the signal. If you're looking for a "brick wall"
limiter, the switch should be set for Hard knee to stop any transients from
slipping through without affecting lower level signals. Lower Ratio levels
may require a hard knee setting so that the compression slope isn't too
narrow and you loose some of the compressive "punch."

Common compressor settings

Vocal Limiting - Vocalists tend to be one of the most dynamic recording
challenges in any studio or stage. Even though a singer may go from a
whisper to a scream during the course of a song, it's the engineer's job to
keep the vocal's level in line with the rest of the ensemble. You can do
this by setting the compressor with a high ratio and a high threshold. This
way, softer sections will go by uncompressed, and louder peaks will be kept
under control.
Threshold set so that the loudest sections get around -6 of reduction
(usually around 3 o'clock)
Ratio set for 6:1 Hardknee OverEasy® switch engaged
The Threshold should be set so that loud sections get compressed around 6dB
and quiet passages get no compression.

Vocal Compression and Spoken Word - In other cases, you may want to compress
the entire dynamic range of a vocal. This is typical of pop vocals and
voiceovers for radio commercials. Whenever there is signal, there is some
compression taking place; just barely on the soft passages, and up to 12 dB
of reduction during loud passages.
Threshold set so one Gain Reduction Led (-1 dB) lights during the softest
passages with signal (usually around 11 o'clock)
Ratio set for 2:1
Attack set fairly fast (8-9 o'clock)
Release set between 10 and 12 o'clock
Raise output to compensate for gain reduction

Drums - Engineers often compress drum tracks just to get a nice punchy sound
in the mix. The settings below sound good on a rock snare drum:
Threshold set so that all drum hits are compressed (around -3dB)
Ratio set for 4:1
Hardknee OverEasy® engaged
Attack set around 8 o'clock
Release set around 9 o'clock
By turning the threshold down even more, you can "squash" the snare drum as
much as you want. Turn the attack up (longer) to get more stick out of the
snare drum, and turn it down for a synth pop slap.

Bass - Since bass guitar forms the foundation of most Rock and Jazz music,
it's important that the level of the Bass doesn't jump around in the mix.
Also, adding compression to bass tracks (or almost anything else) can make
it "punchier," generally a good thing in rock tunes. Try the settings below
on a rock bass track:
Threshold set so only the peaks are compressed (around 0dB)
Ratio set for 4:1
Hardknee engaged
Attack set around 9 o'clock
Release set around 10 o'clock

Electric Guitar - Funky rhythm guitar parts love compression. Not only does
it make the part punch out the mix better, it evens out the volume of the
muted strums. The following setting, with its low threshold and high ratio,
gives you lots of compression for punching up a funky rhythm guitar part:
Threshold set for constant compression (around -3dB)
Ratio set for 6:1
Hardknee OverEasy® engaged
Experiment with turning the Threshold up or down for a thinner or chunkier
tone.

De-Essing - Occasionally when recording vocals, the letter "s" seems to jump
out louder than the rest of the part. This is because sibilant letters,
especially the letter s, have more high frequency energy than other letters.
This can cause tape recorders or other components to distort, even though
the level may not seem very loud. Moving the microphone can sometimes
eliminate this "sibilance," but often a de-esser is required.
Many compressors allow you to perform de-essing on a track by using a
sidechain. By placing an equalizer in the sidechain, you can set the
compressor so that only certain frequency ranges trigger the unit to start
compressing.
The trick is to set the EQ to cut all frequencies except for the sibilant
range, between 3- 6kHz.Set the compressor like this:
Threshold set around 0dB
Ratio set for 6:1
Hardknee active
Attack set at minimum
Release set around 8 o'clock
The Threshold should be set so that an "s" triggers about -3 to -6 dB of
compression. If other sounds are triggering the compressor, you might need
to adjust the EQ cut-off frequencies.

Ducking - Ducking is often used when doing voiceovers. It allows background
music to automatically be turned down whenever an external source, such as
an announcer's voice, begins to speak. You can also use ducking to have one
instrument push the other out of the way, such as the bass guitar ducking
every time the kick drum hits. To make the compressor into a ducker, plug
the source into the inputs and plug the trigger into the sidechain. The
Sidechain return isn't used in this example. In the example below, the sound
of a radio announcer's voice will automatically turn the music down when he
speaks and it will slowly fade back in after he stops:
Set the compressor controls like this:
Threshold set for +3dB (around 3 o'clock)
Ratio set for 6:1
Hardknee OverEasy® in
Attack set around 9 o'clock
Release set around 2 o'clock
Plug the announcer's Mic into the mixer, and feed that Mic to the compressor
's Sidechain in. When the announcer speaks, the music will duck down (turns
the ratio up to duck it even lower). When he finishes speaking, the music
will fade back up at a rate set by the Release knob.

Pumping and Breathing - When a compressor is making large changes to the
input signal (10 to 12 dB or more); the noise floor will also rise and fall
with the signal level. When this noise signal rises and falls drastically
between signals, such as a heavily compressed, noisy drum track, you might
hear the noise level "breathing" between drum hits. One solution to this
breathing problem is to turn up there release time. This way, the noise
floor won't have time to rise between drum hits.

However, if the Release time is too long, lower level signals after the peak
will be lost as the compressor slowly stops reducing gain. this is called
"pumping" as the lower level signals (noise included) slowly fade back up to
their normal signal level. the secret to avoiding these problems is to
achieve a balanced release time on the input signal.

~ 30 ~

Nice primer. I looked around for something like this when I started in audio,
but I do believe one also has to simply work with compressors, make all the
mistakes, until a sense of subtlety and artistry is developed.