Ah, sidechains.
So much potential for creative fun. So much potential for adding groove. For keeping instruments out of each others way. For totally rampant overuse throughout the last decade at least.
With that last point I'm referring to sidechain compression, which has been used to create heavy pumping effects in countless songs over the past while. It's use has become so prevalent it simply seems normal to me now.
I've used sidechain compression for a while now. I like the way it opens up space for my kicks and enhances groove. For the most part, I don't go too wild with it, though from time to time I'll get into some heavy pumping pad action.
Anyway, today's post is about sidechain processing - compressors, gates and expanders. Between these three processors, and a variety of input signals, lies a whole bunch of creative fun.
I'll just touch briefly on what sidechaining means and how it is typically set up.
Normally, when we run a sound through a dynamics processor, it reacts to the sound running through it - a compressor will turn the level down when the sound goes above the threshold, a gate opens when the sound goes above the threshold. With sidechaining, the processor doesn't care what's going on with the sound it's working on. It reacts to a separate signal, coming in via the sidechain input. The input signal (or "trigger") is typically set up on a separate channel in our sequencer, and is routed into the sidechain input. Depending on how we're using it, we may turn the trigger's level all the way down and send the signal to the sidechain input pre-fader. This means that it doesn't matter how much we move the trigger's fader, including turning it all the way down - the signal will be sent to the sidechain input at a constant level.
I'll go through some examples of how sidechaining can be used:
Sidechain compressing a bass to a kick: Low frequencies can quickly get out of control when both a kick and a low bass play at the same time. Sending the kick into a compressor acting on the bass means the bass will clear out of the way every time the kick plays. Problem solved! For the input trigger, there are a couple of options. We could use the actual kick we're using in the song, and send it post-fader. This means that only as much compression as required will occur - if the kick fades in or out at some point, the compression will gradually increase/decrease as required. However, sometimes we want to apply a steady, consistent pump to our bass, regardless of what's happening with the kick we can hear. An option is to put a copy of the kick on a separate track, program a steady beat, turn the level down to 0 and send it to the sidechain compressor. This way the bass will pump consistently throughout the song. Yet another option is to use a different trigger - kick drums often have long boomy tails, which can compress more of the bass than is actually required. For this reason, I generally use a very short closed hihat sample as the trigger. This allows me to set the timing of the pump very precisely by using the compressor's hold or release parameters.
Sidechain compressing any other sound to a kick: See above. Pads, pumping heavily on the offbeat, are common. Leads can come down very briefly, using a low ratio - hardly noticeable, but helpful for clearing out room for the kick. Percussive loops can groove in time with the kick. Heck, we could go so far as to send every sound except the kick into a group channel and compress the entire group every time the kick hits. Mega pump!
Sidechain compressing guitars/pads to a lead or vocal: This is similar to what radio announcers do when they speak over the top of a song that's playing - they start speaking, the music instantly drops down in volume so we can hear them clearly, then the music rushes back up when they stop speaking. We won't generally apply the effect in anywhere near as extreme a way - it's quite unmusical. But the principle remains - we want the centerpiece of our song to come through clearly - we want to make out every word, hear every note of the melody. So we lightly compress a wall of guitars using a vocal input, or a big pad using a lead input. Done subtly, most listeners won't even notice that it's happening - we've simply found a way to go over 0 dBs in the digital realm, that's all. (please don't take that statement at face value!) For this application, we want to send the vocal/ lead in post-fader, so the other sounds drop back only as much as needed. We'll generally want to use a release slow enough that the sound comes back up naturally when the vocal/ lead stops, rather than rushing up and letting the listener know exactly what was going on - though obviously this is a creative decision for each of us!
Sidechain compressing an instrument's reverb/delay to that instrument: This effect turns the reverb down when the instrument is playing, then brings the reverb back up as the instrument fades away. There are a number of creative applications for this, ranging from very subtle to quite extreme. It can be applied to short room reverbs and large caverns. Likewise, it can be applied to delays. It could potentially be applied to other effects (distortion, chorus and so on), but the effect would be much shorter in nature - it would only apply to the end of each note (unless it's a distortion which feeds back on itself!)
Sidechain gating/expanding a pad with a rhythmic pattern/loop: Using a short, snappy gate, a pad can be chopped into an interesting lead pattern by running a drum loop (or any other rhythmic element) into the gate's sidechain input. If the pad features some nice evolving modulation, this will keep running throughout the chopped up sequence - quite different to actually playing the same sequence and having the modulation start from the same point each time a note is triggered. Rhythmic loops with a wide dynamic range can work better than highly compressed loops - a wide dynamic range means there is a more defined space between "note on" and "note off". Using compressed loops will require finer tuning of the input threshold. A more subtle version of the chopped gated effect can be created by using an expander with a low ratio - the sound will only drop marginally between transients, rather than completely cutting out. Using automation, this also allows us to smoothly transition from unbroken pad chords/notes to heavily chopped chords/notes and back again.
Automating or changing the input trigger: A lot of the time, the automation will need to be reasonably heavy-handed to make an impact on the way the dynamics processor reacts. The input trigger can be heavily EQ'd - if we are punching out a rhythmic pattern using a percussive loop into a gate, an EQ can change the percussive loop so the kick comes through much more prominently (or another area of the loop). A transient designer can be attached to the percussive loop, transitioning it from very short and snappy to more smooth and drawn-out.
There are many more creative possibilities. Whether it's punching parts of sounds out or punching them in, whether it's pushing something away from another sound or locking it to it, or even whether it's a very subtle application which transparently solves an issue, sidechaining is a very handy resource to have at our fingertips.
Have fun, keep making great music!
Fabian
Showing posts with label Compression. Show all posts
Showing posts with label Compression. Show all posts
Thursday, June 2, 2011
Thursday, March 24, 2011
Hearing what a compressor does
Some effects are easier to hear than others. Run a synth lead through a big hall reverb and we'll notice the massive space around it. Grab an EQ and boost 5 kHz and we'll hear the high end definition (or shrillness, depending on the original sound) coming through.
Compression, however, can initially seem like a black box. We know it's supposed to "pump up" our sounds, make them bigger, stronger, louder, better. But how? And what are all those controls and why do some of them do nothing, even though we turn them all the way from one end to the other?
Well, let's start with the basics. A compressor turns the volume down when the incoming signal goes over the threshold we've set. That's all a compressor actually does - it turns the volume down. In practice though, there's a bit more to it - how quickly it turns the volume down, how long it should leave the volume down when the incoming signal drops back below the threshold. How much gain can be added to the overall output signal now that the loudest parts have been turned down.
There is a way to hear what a compressor is doing to the signal, and to set the compressor so it does what we want it to do, in an appealing musical way. This is preferable to choosing minimum attack and release, and maximum ratio, which will make the signal hotter, but may not make our song sound better. It's usually better if the compressor works in the same groove as the music.
First, set the threshold so that the compressor is being triggered with every note/ beat that comes through it. This means we'll be able to clearly hear how the compressor operates on the sound.
Next, set the ratio to maximum. Yes, I just spoke against "maximum-ratio-itis", but here it's just an initial setting, which will allow us to clearly hear the timing of the attack and release parameters.
Next, set the attack and release to minimum. Once again, it's not likely they'll end up there, but these starting positions will allow us to hear clearly what the compressor is doing to our sound.
If the compressor has a "hold" parameter, set it to zero. The makeup gain/ auto makeup isn't important at this point - the final output level can be set when we have the compressor working on the sound the way we want it to.
Okay - the compressor should be heavily compressing the sound as it's playing.
The first parameter we'll adjust is the attack parameter. My main consideration here is whether the sound has an attack transient, and whether I care about the transient. If it's a percussive loop, or a kick drum, I'll almost always slow the attack down so that the attack transient comes through cleanly. If we leave the attack at minimum, the front end of the sound will dull considerably - rather than a sharp, clear front end of the sound which cuts through a mix, we'll be left with something which sounds much more like it's being hit with cardboard. The body of the sound will be fine, but the clear attack transient will disappear. How slow the attack should be is totally dependent on the sound and how we want the attack transient to sound. For myself, with some lead sounds I'm happy to lose the attack entirely. For pads, which often have a very slow attack phase, I'm happy to leave the compressor's attack at minimum.
Now that we have the attack where we want it, our focus can turn to the release parameter. My main consideration here is whether the sound is rhythmic and needs to fit into the groove of the song. Pads aren't rhythmic - the release can be left as fast as possible. Drum loops are very rhythmic, and will very likely work against the song's groove if the release is at minimum. We want the compressor to move up and down in volume in time with the song, rather than in some arbitrary way independent of the song's natural groove. So when setting the release parameter, start bobbing your head in time with the song, or get up and dance, or whatever, and see where the release parameter best enhances the groove. This is important - this is the difference between whether or not people feel like moving their bodies to our song!
Right, so we have the compressor working on the part of the sound we want it to work on, and the level rising back up in time with our song. Now it's time to adjust how much compression we want to apply. The last two controls work in tandem - a very low threshold, where the compressor is working on the full range of the sound, combined with quite a low ratio (say 1.20:1) will give us a large reduction in dynamic range. So will a much higher threshold which only catches the loudest parts of the sound, combined with a much higher ratio (say 8:1). As with so many things, it comes down to personal choice - we're artists, after all! We may want to even out the instrument and make it stronger and steadier in the mix, but to still retain enough dynamic range to allow the instrument to play more quietly (say, during a trance breakdown) for contrast. We may want to slam it hard for a very "in your face" sound. Do we want to compress just the spikes, or the full range of the sound? I'm not going to attempt to answer these questions, since we're all going to answer them differently.
I know some people like to deal in approximate settings, so here are some typical starting settings I use for sounds in trance. Where they end up is totally dependent on the sound and the song:
Kick - I generally don't compress kicks. The samples I use are loud and punchy enough for me. If anything, I use a transient designer to heighten the attack portion of the sound.
Low/ Main Bass - I often use a vintage emulation compressor on my main bass, though my bass is usually very solid in the mix without it. I use this compressor more for the tone than the compression it imparts. More often, I'll run the bass through a sidechain compressor to briefly push it out of the way of the kick. This pushes the bass down quite heavily on each kick, with the bass rushing back up as fast as sounds good, musically.
Mid/High Basses - again, these usually receive a sidechain compression treatment similar to the low/ main bass. When I do compress them with standard compression, I use reasonably fast attack (say 10-20ms), reasonably fast release (20-40ms) and aim for 3-6 dBs of gain reduction. It depends how "weighty" I want the sound to be in relation to the other sounds around it. This is a big change to my early days, when my thoughts were "everything needs to be ultra loud and in your face". Now I understand that the more in-your-face one sound is, the less in-your-face all the other sounds become.
Pads - If the pads feature a fair bit of modulation and drift in and out a bit much for my liking, I'll compress them in a similar way to the synth group (see below) - in short - very low threshold and ratio, minimum attack and release.
Synths - I often compress these twice, firstly on their own, and secondly via a sidechain compressor. The settings I use for the first compressor operating on the synth group are as follows: a very low threshold, so the compressor is always operating on the sound. Then quite a low ratio (say 1.50:1), which over the range of operation still amounts to a decent amount of gain reduction, just spread over a large dynamic range. The attack and release parameters are set to minimum. Since the compressor is operating non-stop, I assume that having these set as fast as possible means the compressor makes adjustments as fast as possible. This treatment brings the entire dynamic range of the synths together smoothly, without any non-linear points where the compression is coming in and out.
Drums - I generally don't compress Hi Hats and Claps/Snares.
Percussive Loops - As for other elements, these are sometimes compressed using a sidechain compressor, using a release setting which makes the loop groove nicely around the kick drum.
Sound Effects - I generally don't compress these. If I do, it's generally via a sidechain compressor to add the pumping effect to up- and down-lifter effects.
Overall mix - As for the pads/ synth group, a very low threshold and extremely low ratio - lower than for individual groups - usually only 1.10:1, so hardly noticeable. However, it does mean I don't need to work the master limiter quite as hard.
So that's my approach to compression in a nutshell. I hope you've found it useful!
Fabian
Compression, however, can initially seem like a black box. We know it's supposed to "pump up" our sounds, make them bigger, stronger, louder, better. But how? And what are all those controls and why do some of them do nothing, even though we turn them all the way from one end to the other?
Well, let's start with the basics. A compressor turns the volume down when the incoming signal goes over the threshold we've set. That's all a compressor actually does - it turns the volume down. In practice though, there's a bit more to it - how quickly it turns the volume down, how long it should leave the volume down when the incoming signal drops back below the threshold. How much gain can be added to the overall output signal now that the loudest parts have been turned down.
There is a way to hear what a compressor is doing to the signal, and to set the compressor so it does what we want it to do, in an appealing musical way. This is preferable to choosing minimum attack and release, and maximum ratio, which will make the signal hotter, but may not make our song sound better. It's usually better if the compressor works in the same groove as the music.
First, set the threshold so that the compressor is being triggered with every note/ beat that comes through it. This means we'll be able to clearly hear how the compressor operates on the sound.
Next, set the ratio to maximum. Yes, I just spoke against "maximum-ratio-itis", but here it's just an initial setting, which will allow us to clearly hear the timing of the attack and release parameters.
Next, set the attack and release to minimum. Once again, it's not likely they'll end up there, but these starting positions will allow us to hear clearly what the compressor is doing to our sound.
If the compressor has a "hold" parameter, set it to zero. The makeup gain/ auto makeup isn't important at this point - the final output level can be set when we have the compressor working on the sound the way we want it to.
Okay - the compressor should be heavily compressing the sound as it's playing.
The first parameter we'll adjust is the attack parameter. My main consideration here is whether the sound has an attack transient, and whether I care about the transient. If it's a percussive loop, or a kick drum, I'll almost always slow the attack down so that the attack transient comes through cleanly. If we leave the attack at minimum, the front end of the sound will dull considerably - rather than a sharp, clear front end of the sound which cuts through a mix, we'll be left with something which sounds much more like it's being hit with cardboard. The body of the sound will be fine, but the clear attack transient will disappear. How slow the attack should be is totally dependent on the sound and how we want the attack transient to sound. For myself, with some lead sounds I'm happy to lose the attack entirely. For pads, which often have a very slow attack phase, I'm happy to leave the compressor's attack at minimum.
Now that we have the attack where we want it, our focus can turn to the release parameter. My main consideration here is whether the sound is rhythmic and needs to fit into the groove of the song. Pads aren't rhythmic - the release can be left as fast as possible. Drum loops are very rhythmic, and will very likely work against the song's groove if the release is at minimum. We want the compressor to move up and down in volume in time with the song, rather than in some arbitrary way independent of the song's natural groove. So when setting the release parameter, start bobbing your head in time with the song, or get up and dance, or whatever, and see where the release parameter best enhances the groove. This is important - this is the difference between whether or not people feel like moving their bodies to our song!
Right, so we have the compressor working on the part of the sound we want it to work on, and the level rising back up in time with our song. Now it's time to adjust how much compression we want to apply. The last two controls work in tandem - a very low threshold, where the compressor is working on the full range of the sound, combined with quite a low ratio (say 1.20:1) will give us a large reduction in dynamic range. So will a much higher threshold which only catches the loudest parts of the sound, combined with a much higher ratio (say 8:1). As with so many things, it comes down to personal choice - we're artists, after all! We may want to even out the instrument and make it stronger and steadier in the mix, but to still retain enough dynamic range to allow the instrument to play more quietly (say, during a trance breakdown) for contrast. We may want to slam it hard for a very "in your face" sound. Do we want to compress just the spikes, or the full range of the sound? I'm not going to attempt to answer these questions, since we're all going to answer them differently.
I know some people like to deal in approximate settings, so here are some typical starting settings I use for sounds in trance. Where they end up is totally dependent on the sound and the song:
Kick - I generally don't compress kicks. The samples I use are loud and punchy enough for me. If anything, I use a transient designer to heighten the attack portion of the sound.
Low/ Main Bass - I often use a vintage emulation compressor on my main bass, though my bass is usually very solid in the mix without it. I use this compressor more for the tone than the compression it imparts. More often, I'll run the bass through a sidechain compressor to briefly push it out of the way of the kick. This pushes the bass down quite heavily on each kick, with the bass rushing back up as fast as sounds good, musically.
Mid/High Basses - again, these usually receive a sidechain compression treatment similar to the low/ main bass. When I do compress them with standard compression, I use reasonably fast attack (say 10-20ms), reasonably fast release (20-40ms) and aim for 3-6 dBs of gain reduction. It depends how "weighty" I want the sound to be in relation to the other sounds around it. This is a big change to my early days, when my thoughts were "everything needs to be ultra loud and in your face". Now I understand that the more in-your-face one sound is, the less in-your-face all the other sounds become.
Pads - If the pads feature a fair bit of modulation and drift in and out a bit much for my liking, I'll compress them in a similar way to the synth group (see below) - in short - very low threshold and ratio, minimum attack and release.
Synths - I often compress these twice, firstly on their own, and secondly via a sidechain compressor. The settings I use for the first compressor operating on the synth group are as follows: a very low threshold, so the compressor is always operating on the sound. Then quite a low ratio (say 1.50:1), which over the range of operation still amounts to a decent amount of gain reduction, just spread over a large dynamic range. The attack and release parameters are set to minimum. Since the compressor is operating non-stop, I assume that having these set as fast as possible means the compressor makes adjustments as fast as possible. This treatment brings the entire dynamic range of the synths together smoothly, without any non-linear points where the compression is coming in and out.
Drums - I generally don't compress Hi Hats and Claps/Snares.
Percussive Loops - As for other elements, these are sometimes compressed using a sidechain compressor, using a release setting which makes the loop groove nicely around the kick drum.
Sound Effects - I generally don't compress these. If I do, it's generally via a sidechain compressor to add the pumping effect to up- and down-lifter effects.
Overall mix - As for the pads/ synth group, a very low threshold and extremely low ratio - lower than for individual groups - usually only 1.10:1, so hardly noticeable. However, it does mean I don't need to work the master limiter quite as hard.
So that's my approach to compression in a nutshell. I hope you've found it useful!
Fabian
Subscribe to:
Posts (Atom)