Sound Workshop
Technique : Effects / Processing
With the powerful automation available in
modern DAWs, it is easy to overlook their side-chaining features. But
what value does this technique have in the modern project studio?
 |
While software plug-in emulations of compressors and
gates may include options for external side-chain access, or for
filtering the side-chain signal, it seems that many users are ignoring
these features because they're not entirely sure what they are for.
Essentially, they do the same job as their hardware counterparts and
this short feature aims to explain exactly what a side-chain is and how
external access and filtering may be used.
Both gates and compressors normally work by
monitoring the amplitude or intensity of their own inputs, so that they
can detect whether the level is above or below the threshold value
chosen by the user. While some classic hardware compressors use an
alternative system, where the threshold is fixed and the user instead
sets the amount of compression by adjusting the input drive level, the
outcome is much the same: if the signal rises above the threshold, the
compressor or gate responds.
The part of the circuitry that monitors the input
level is known as the side-chain, and it controls that part of the
circuitry that adjusts the gain of the main signal path. In previous
articles I've likened this to an engineer who listens to the sound level
coming over his monitors and then makes adjustments using a mixer fader
when the sound gets too loud.
In most situations the side-chain monitors the input
level, although there are some designs, particularly limiters, that
monitor the output level instead. Although it is significant to the
technical operation of the device, this circuit topology detail doesn't
greatly affect the way the user sets up the compressor.
One very simple way of modifying the behaviour of a
dynamics processor is to apply high- or low-cut EQ to the signal before
it reaches the side-chain. This doesn't affect the tonality of the main
signal path but it does affect the way in which the side-chain responds
to what it hears. In the case of a gate, side-chain filtering is often
built-in and is used to make the gate less sensitive to the extremes of
the audio spectrum so as to prevent, or at any rate reduce the
likelihood of, false triggering from sources such as microphone spill.
The best way to explain this process is to consider an example.
Let's assume that we have a microphone on a tom-tom
that is part of a drum kit. We have set up the gate to get rid of the
low-level ringing that often occurs as a result of sympathetic resonance
(usually when you hit the kick drum). Because this resonance is at a
fairly low level, we can set the gate threshold so that the gate closes
shortly after the tom has been hit. So far so good. But if you have a
cymbal near the tom and hit it hard, the amount of spill into the tom
mic might be enough to cause the gate to open. By putting a high-cut
filter before the side-chain, we can EQ some of that cymbal sound out of
what the side-chain 'hears' so that it is less likely to cause the gate
to open in error. Yet, when the gate opens, you can hear the tom-tom
without the EQ.
Similarly, if the bass drum was being picked up in
the snare mic and causing the snare gate to open unnecessarily, you
could put a low-cut filter before the side-chain input to reduce the
risk. Drawmer gates have both high- and low-cut filters with controls to
vary their frequencies, so that the gate essentially triggers from the
frequency range bounded by the two filters. Material falling below the
cutoff frequency of the low-cut filter or above the cutoff frequency of
the high-cut filter is largely ignored. This filter arrangement has
since been adopted by a number of other hardware and plug-in gate
manufacturers.
To set up the filters most effectively, you need
some way to hear what the side-chain is hearing and this is usually
provided by a 'side-chain listen' or 'key listen' switch. When this
switch is active, you hear the filtered version of the signal, enabling
you to adjust the filters to pass as much as possible of the wanted
sound and as little as possible of the spill that is causing you
problems. Once the filters are set, you can turn off the key listen and
adjust the gate threshold and release time to get the best results. It
is important to realise that the side-chain signal doesn't have to sound
'nice', as nobody is going to hear it. What you are trying to do is
emphasise the amplitude difference between the strongest part of the
wanted signal and the unwanted ones, and that usually demands very
severe and harsh EQ settings.
In the case of drums, don't worry if the closing of
the gate sounds a little unnatural on its own when soloed or if the
occasional drum or cymbal breaks through, because these artifacts tend
to get completely hidden when you add the drum overhead mics (which
should not be gated). What you are trying to achieve is an improvement
in separation and control. Although the individual elements may not
sound perfect in isolation, the end result should sound much better than
if you chose not to use a gate at all.
External Gate Keying
Rather than feed the gate's side-chain from its own
input, you may find that you have an external side-chain input that
allows you to control the gate from a separate signal source (usually
another audio track in the case of a plug-in). One way to visualise this
is to imagine an engineer adjusting a fader according to the level of a
track he's hearing over the monitors but the fader is in fact
controlling a different track altogether (see diagrams, right).
The diagram illustrates two alternative
configurations for side-chain control. In the top diagram, the gain
control is determined by the primary audio input signal, whereas in the
lower one, a secondary signal is routed to the side-chain and therefore
controls the gain of the other.
|
One of the oldest examples of how this might be
useful is to place the gate in a track that has an over-busy bass guitar
part, then trigger the gate's side-chain from the kick-drum track. This
forces the gate to open only when a kick drum is present after which it
closes at the rate you set using the release control. As a result, the
only bass guitar you hear is that which plays at the same time as the
kick drum. The rest gets gated out. How natural this sounds varies,
depending on how the bass part was played originally, but this trick can
tighten up an otherwise messy bass part. You can also feed in rhythmic
signals to produce controlled 'chopping' effects: for example, breaking
up a synth pad in order to create rhythmic bursts.
Yet another trick possible using external keying is
ducking, although the gate must have an attenuation control that allows
you to input negative as well as positive values for this to work.
Ducking is traditionally done using compressors, but a number of modern
hardware and plug-in gates have the necessary features to allow them to
function as very efficient duckers. Normally a gate has an attenuation
control that determines by how much the signal level is reduced when the
gate is closed. If the attenuation can be set to a negative value, as
it can in Logic's gate, then the signal level actually
increases when the gate is closed. What's more, you can dial in by
exactly how many dBs you want the level to increase when the side-chain
input signal falls below the gate's threshold.
A good example of how this feature might be useful
is placing the gate in a busy guitar track and then triggering the
side-chain from the lead vocal. You can then adjust the gate so that
when the vocal is present, the gate is triggered and the signal level
passes through at unity gain. When the vocal is absent, the gate closes
and the guitar part is boosted by however many decibels of negative
attenuation you have set on the gate controls.
The reason for doing this is so that you can have
the guitar part a couple of dBs louder in the absence of vocals. Then,
when the vocals are present, the guitar level drops back a little to
allow the vocals to be heard more clearly. The technique is exactly the
same as the one used by radio DJs, where a ducker is used to drop the
level of the record whenever the DJ speaks. This is a useful way of
keeping the vocals audible in a busy mix without having to automate the
instrument levels manually.
Compressors
We can access the side-chain in exactly the same way
with many compressors and, as you're probably aware, compressors can
also be used to perform ducking in conjunction with a side-chain input
from another track. However, the end result of filtering the side-chain
in a compressor is quite different to what happens in a gate, even
though the object of the exercise is still to make the device more or
less sensitive to specific frequencies.
For example, to use a compressor for de-essing, we
simply insert a peaking (band-pass) filter into the side-chain input
path and boost the signal in the region where sibilance occurs
(typically 4-8kHz). If we were to boost that range by 10dB, then the
compressor would be 10dB more sensitive to those frequencies and apply
gain reduction whenever those frequencies were loud enough to push the
signal level over the threshold. Of course, a simple de-esser like this
reduces the gain of the entire audio signal whenever there is enough
level in the sibilance range to trigger the threshold, which means that
over-use can lead to the vocals taking on a rather unnatural lisping
quality. However, it still provides a good example of this principle.
Logic's Compressor plug-in is being used
for 'ducking' in this screenshot. The compressor is inserted on the
first track (a backing submix) but its actions are controlled by the
second track (vocal). When the vocal signal exceeds the threshold, the
submix is automatically compressed and reduced in volume. The submix
itself has no impact on the operation of the compressor. Note that, to
compensate for the attack time of the compressor, the vocal track has
been slightly advanced ahead of the submix.
|
By setting the peaking filter to around 40Hz, the
compressor can be made most sensitive to very low frequencies, which may
be useful when trying to reduce the subjective impact of popping on a
vocal track. Although this won't eliminate pops, it will reduce them in
level.
While most software plug-in compressors don't allow
you to insert your own filters into the side-chain input, there are
often dedicated de-esser plug-ins available that combine the necessary
compressor and side-chain filters. With a suitably equipped hardware
compressor, you can patch any type of equaliser into the side-chain
input using conventional patch cables. Note that some hardware
compressors include a side-chain insert point, which makes it even
easier to patch in your equalisers.
One workaround for creating a de-esser, using a
compressor plug-in with an external side-chain input, might be to copy
the audio being processed to another audio track, and set up the
compressor to access that as its side-chain source. Then you use a
filter plug-in on your newly created side-chain control track to locate
the problem area, such as sibilance, and boost those frequencies as much
as possible, taking care not to allow the levels to clip. You can use
your bounce function to make this EQ change permanent if your DAW takes
its side-chain feed before any insert plug-ins added to the control
track. Now the EQ'd track will control the compressor so that it is most
sensitive to the region you have boosted.
Compressors are also widely used for ducking, but in
my view they are not as predictable in this role as gates. Why? The
amount of gain reduction applied by a compressor depends on the ratio
you set, and on the amount by which the input signal rises above the
threshold. If you control the ducking process using a vocal track as the
side-chain input, the most gain reduction will be applied when the
vocal is loudest, when what you really want is either more gain
reduction to drop the backing parts when the vocal is quieter or a
system that always drops the level by the same amount once the vocal
track rises above the threshold. With a gate you can be very precise as
to how much gain reduction is applied but using a compressor it will
always vary to some extent. However, the higher you set the ratio, the
less the amount of gain reduction will be affected by the absolute level
of the side-chain signal, as long as it is high enough to cross the
threshold. For this reason, a compressor with a ratio control that goes
right up to hard limiting is the best choice for ducking.
A potential problem when using either a gate or
compressor for ducking is that the 'duck' can't happen until
fractionally after the controlling signal has started. This can be fixed
in a DAW by advancing the control track in time so that the duck
happens fractionally early, and setting the release and hold times so
that the 'un-ducking' rate is appropriate. You might also consider using
plug-ins that have an inbuilt look-ahead facility, although most will
probably not look ahead far enough for this purpose. Of course,
level-automation is used to do this manually and more precisely in many
cases, so duckers are not quite so indispensable as perhaps they once
were.
Most of the time, compressors and gates are set up
to process their own input signals, so all you have to do is patch them
into an insert point. I hope this brief introduction to side-chains will
inspire you to explore some of the more esoteric applications of gates
and compressors, whether hardware or software.
No comments:
Post a Comment