Reference Manual
CHAPTER 21. LIVE AUDIO EFFECT REFERENCE 291
A compressor can only react to an input signal once it occurs. Since it also needs to apply
an attack/release envelope, the compression is always a bit too late. A digital compressor
can solve this problem by simply delaying the input signal a little bit. Compressor offers
three different Lookahead times: zero ms, one ms and ten ms. The results may sound pretty
different depending on this setting.
Compressor can be used in three different EF (envelope follower) modes. With Peak se-
lected, Compressor reacts to short peaks within a signal. RMS mode causes Compressor
to be less sensitive to very short peaks and compress only when the incoming level has
exceeded the threshold for a slightly longer time. Opto mode results in a non-linear release
curve. Specically, the release behavior is faster initially, and slows down as the gain reduc-
tion approaches zero. So which type should you use? There's no right answer, of course,
but there are some common uses for each mode. Peak mode is more aggressive and pre-
cise, and so works well for limiting tasks where you need to ensure that there are absolutely
no signals over the set threshold. RMS is closer to how people actually perceive loudness
and is usually considered more musical. Opto mode, because of its non-linear release
time is often considered smooth and natural sounding. Opto compressors are commonly
used on vocals, bass and electric guitar. But as always, when it comes to compression, trust
your ears and not the meters!
In addition to the EF modes, Compressor offers three Model types that offer further options
for how the device measures and responds to signal levels. FF1 and FF2 are both feed-
forward models, which analyze the loudness of the incoming signal. These are the models
used in the former Compressor I and Compressor II devices, respectively. FB is a feedback
model, which analyzes the output of the device and then self-adjusts its compression be-
havior. Because feedback compressors analyze signals that have already been compressed,
their Attack and Release parameters are a bit less precise, and act more like suggestions to
the compressor. But feedback compression generally results in a much smoother sound,
with less overall gain reduction but also less potential for distortion artifacts. The classic
analog compressor models that are so highly sought after in hardware-based studios are
generally all feedback models. Note that, for reasons of quantum physics, Lookahead and
external sidechaining are disabled when using the FB model; Ableton's engineers are hard
at work developing code that will allow our software to predict the future, but we don't
anticipate having this available until at least the next major release.