Hi, and welcome to this beginner’s guide
to bus processing. Lets start by defining the term: a “Bus”
simply adds multiple signals together. The most important bus in your project is
the master bus, otherwise known as the mix bus, where all your individual channels are
summed together to create one stereo signal. If you add processing to this master channel,
you’re processing the entire mix. We also use buses when we create send effects. If I send a few different channels to a reverb
effect… these channels all get mixed together, before being processed by an all-wet reverb
effect… and the resulting reverb-only signal can then be blended with the dry signals. Technically this is bus processing, as the
signals are added together at a bus, and then processed by the reverb. But this kind of effect send setup isn’t
what we’re usually referring to when we use the term bus processing, as in this case
the dry signal still just goes directly to the mix bus, and if I turn down the reverb
channel we only lose the reverb effect. The term “bus processing” more usually
refers to the use of subgroups. In this case the dry signals don’t go directly
to the mix bus: first they get mixed together with a group of other related channels, and
then that subgroup is routed to the mix bus. The method you use to achieve this will vary
depending on the DAW you’re using: in Reaper I can create a folder track, in this case
containing all my drum channels, and the folder track will automatically become a subgroup. Other DAWs might use a more traditional hardware
mixer paradigm, such as Harrison Mixbus: in this case I need to change the output routing
for my drum channels, so they all go to the drum bus instead of direct to mix. This allows us to adjust the level of the
whole drum kit using just a single fader… but also allows us to process the whole drum
kit by loading plugins onto the subgroup channel. Lets consider a few of the most common options. If I load an EQ onto the subgroup I can EQ
all the drums together. These classic drum machine samples have bags
of character, but they all benefit from a gentle cut in the lower mids to reduce the
boxy character… and another to stop this lower region conflicting with the bass part…
and as the kick drum is the only element with significant content in the low sub bass, I
can shape the low end of the kick in any way I like without really affecting anything else. In this case, EQing the group gives exactly
the same results as if I EQed every channel individually using the same settings... But using just a single instance on the group
is much more efficient. A single EQ instance uses fewer of your processor
cycles than would multiple instances, but this is not so important as any modern processor
can handle many instances of Pro-Q3 without breaking a sweat. More importantly, it allows you to work more
quickly, which makes you more productive, and probably leads to better results: the
longer you spend listening to a mix in progress the more you lose your perspective, and the
harder it becomes to gauge how your mix will sound to a fresh pair of ears. This approach doesn’t preclude also EQing
individual channels. For example, I might try crisping up the sound
with a boost around 8KHz… but then cut back that same region for the hi hat channel to
stop it getting too bright... and this is still more efficient than loading and adjusting
a separate EQ instance for every drum channel. Starting with broad brush EQ changes applied
to subgroups can be a very effective way to get a good basic mix together quickly, which
can then be refined with tweaks to individual channels if needed, and this approach is often
referred to as “top down mixing”. The reason a single EQ on a subgroup can be
equivalent to EQing each individual channel is because a clean EQ such as Pro-Q3 is totally
linear. In DSP terms this means it behaves the same
way regardless of the incoming signal level, and hence EQing each channel before adding
them together is the same as adding them together first, then EQing the result. This ceases to be true as soon as you start
to use Pro-Q3's dynamic features however. Now we have a non linearity, because this
dynamic band is reacting to the input signal and compressing the dynamic range for that
region. If I ran multiple instances of this setting
on individual channels, each instance would see a different input signal, resulting in
different dynamic behaviour. Ok now lets try compression. Here’s how it sounds if I compress each
element individually… the kick and snare now sound punchier and more focussed… and
we’ve lost most of the dynamics in the hi hat part. Now lets try compressing them all on the subgroup
instead: I can achieve a similar extra attack and focus for the kick and snare parts…
but the hi hats aren’t hitting the threshold at all with this setting, so we’re not losing
the accents of the original part. But because the compressor gain reduction
is applied to the mixed subgroup, the gain reduction applied to kick and snare hits now
also affects the hi hat. Changing the release time doesn’t only change
the decay of the kick and snare hits: it now also changes the way the hi hats ride up in
level in the pauses between kick and snare hits, and can therefore affect the groove
of the drums. Compression is non linear by definition: it
reacts to, and shapes the dynamics of the incoming signal. So unlike EQ, compressing a subgroup is never
the same as compressing the individual elements separately. Here’s a subgroup with all my guitar and
synth parts: when I compress them all together they start to interact, with loud events from
one part ducking all the other parts in level, and this interaction between the parts is
often called “glue”. The glueing effect is strongest when you compress
the entire mix on the mix bus, but glueing together just a subgroup can also be a very
useful approach. Watch my Beginner’s Guide to Compression
for a more detailed look at this effect. Saturation is also a non-linear process: turning
up the input level results in more saturation, and more added harmonics, while turning it
down makes the sound cleaner. And like compression, saturation will behave
differently when applied to a whole subgroup instead of each individual part. Each part will gain extra harmonics just as
it would if processed alone, but as well as this we’ll also get intermodulation, as
each part changes the way all the other parts hit the non linear distortion stage. Intermodulation quickly starts to sound nasty
if there’s too much of it, so you’ll usually need to keep the saturation pretty subtle
when processing subgroups or a full mix. But in small amounts it adds its own subtle
kind of glueing effect, helping the parts to gel together better. For my money this is the key to the elusive
analogue sound thats so sought after: small amounts of saturation on buses can help to
create that expansive, warm yet clear sound characteristic of many classic analogue recordings…
while too much of it can recreate the muddy, cluttered, or abrasive character of many forgotten
bad recordings from the same era. Check out my Beginner’s Guide to Distortion
and Beginner’s Guide to Saturation videos if you want to know more about intermodulation. Putting them together. In a real world mix we’re likely to be using
EQ and compression and saturation, which complicates things a little. If the EQ on your subgroup comes before any
non linearities such as compression or saturation, then it will still be equivalent to the same
EQ setting applied to all the individual channels in that group, so long as those individual
channel EQs all come after any compressors or saturators on those channels. But if the individual channel EQs are running
before the non linearities… or if the subgroup EQ is running after them… the two will no
longer be equivalent, as EQ before a non linearity isn’t the same as EQ after a non linearity. So what about more creative effects? Obviously a filter sweep is more dramatic
in effect when opapplied to the whole mix instead of just a single element… But equally you could apply creative effects
to a smaller subset of parts, like bass, keys and guitar for example. As complex as this filter setup seems, it
is in fact almost entirely linear: the LFO and step sequencer modulation doesn't care
about input levels, and would still behave the same way if I ran multiple instances on
the individual channel instead. Further to the right we find an envelope follower,
which definitely does depend on input levels… but if we look closer we see that it's set
to listen to the sidechain input, to which I've routed the kick drum, so the resonance
for the lower filter is turned down on every kick drum. Assuming you set up the same sidechain routing
for each channel this would again behave the same way running as multiple separate instances. In fact the only real non linearities in this
preset are within the filters themselves. If I switched both filters to the Clean type
then this preset, as complex as it seems, would be just as linear as a static EQ setting
in Pro-Q3. This Timeless preset is similar: the step
sequencer modulation of the dry wet mix would be the same if applied to each individual
channel… even the delay time modulation would be the same… but I'm again using filter
models which include saturation, and more importantly, I'm modulating them with an envelope
follower, and this time it's set to listen to the main inputs rather than the sidechain. This preset is therefore definitely non linear,
and wouldn't sound the same if applied to each channel separately. Finally here's a Saturn preset, which modulates
the Drive parameter for the Destroy type of distortion. This includes some clipping for very loud
signals, and bit crushing which mostly affects quieter signals, but the most significant
part of the sound is the samplerate reduction, which doesn't care about input level at all. However, the modulation in this preset has
a strong random element, provided by the third XLFO: if you ran individual instances for
each channel, each would have different random values, and the result would be very different. If I put them all together… the individual
parts get glitched up and modulated together, creating something altogether new… In this case the resonance of the Volcano
high pass filter gets a bit overwhelming at times, but rather than turning down the resonance
and losing the self oscillation between kick drums, I'm instead going to use a multi-band
compressor to keep the low frequencies under control. Multi-band compression can also be useful
on buses, to control the extreme low or high ends, or tame harshness in the upper mid range,
or selectively glue parts of the mid range. But we're starting to go beyond beginners
level now, so I'll leave it there. Bye for now, and thanks for watching.
