Why should we care?
Drum processing has to be one of the most thoroughly covered music production topics of all time. Since the earliest eras of electronic music, people have been obsessed with finding new ways to generate, sample, process, and slice percussion. By now, there are a million and one pieces of software dedicated to altering the transients and bodies of a given sound in extremely precise ways — and in turn, we are now capable of shaping drums with more creative freedom than ever before. But today, we look back in time to a technique predating modern production software to unlock new potential in our existing tools. The technique in question we'll refer to as amplitude splitting.
We'll be making an easy-to-use Ableton Live rack that takes advantage of amplitude splitting. You can download it here: download the Ableton rack.
What does it sound like?
Check out the example below to hear one of the many things we can achieve with amplitude splitting. Listen to how rich and defined this sample's space and high end becomes after the processing is applied – as well as the preservation of transients. It almost sounds as though the sample has been excited; however, it's much deeper than that. The trick is enhancing particular dynamics of the sound in a musically pleasing way which preserves important parts of the original signal. It turns this basic drum loop into something ready for modern electronic music and more.
Before amplitude splitting:
After amplitude splitting:
How does it work?
We'll be recreating an old technique known as "latch gating." To do this in the analog world, we would take three microphones and place them at increasing distances from a source. Next, apply a gate to each mic so that a specific volume needs to be reached in order to activate it. The mic closest to the source should have the lowest threshold, the next one should be higher, and the furthest one should be the highest, meaning the third will be the one requiring the loudest signal to become active. Finally, only one mic can be active at once, meaning as one mic turns on, the other two mute.
Now, when a sound is sent through this system, something very loud will trigger the furthest mic, something very quiet will be recorded by the first mic, and everything in between will be from the middle mic. This way, we can split the signal by amplitude and process each part separately.
Don't worry — we're making a simplified software version of this system in Ableton. To make it easy, we'll start with a single split between the loud and quiet parts of the signal. All we need is two gates and a utility in a clever configuration. By the end, our rack should be able to split up any piece of audio in the following way:
Start by creating a single gate and grouping it into an audio effect rack (cmd/ctrl + G). You should map the threshold to a macro. I usually keep all of the knobs at their minimum value for right now; it should look like this:
Now head to the chains section of the rack, where we'll create a parallel channel. Click the existing chain with the gate we made and duplicate it (cmd/ctrl + D). On this chain, add a utility and invert the polarities (L and R buttons in the top left, we'll explain this later). Here's what it looks like:
This last step is the real trick. Select the gate and utility using shift-click and group the effects together using cmd/ctrl + G. On this new group, go to the chain section and create a new parallel channel using right-click > create chain. You can leave it blank because we're done! It should look something like this:
To make the rack a little more user-friendly, I'll rename the first two chains, collapse the gate on the loud signal, and collapse the audio effect rack on the quiet signal. Any processing should come at the end of each chain after the gate/effect rack:
So from here, we have infinite possibilities. One of the first things I noticed after playing with this for a while is how much control we can gain over the space of the sound. I love to take recorded drum kits and listen to the reverberance in the quietest parts of the signal. Something as simple as boosting the gain of the quiet signal can have a dramatic effect — but of course, we can do so much more than that. Distortions, delays, reverbs, transient shapers, frequency shifters — you can add any type of processing to these channels.
We can easily achieve very specific effects with this technique, such as sending only the loudest hits in a drum loop through reverb or preserving the transients and adding delay to the space of the kit. Truly a whole other perspective from which to process your sounds.
In this example, you can hear how only the snare (the loudest part of the kit) has a reverb tail while the rest of the signal remains dry:
In this one, you can hear how adding a phaser to the quiet signal allows you to use the effect subtlety without ruining the transients:
Dial it in
Don't forget to try playing with the gate settings as well. Changing the threshold will determine at what level the split between loud and quiet is, and changing lookahead and attack times will affect the transients. Just make sure that all the values on the gates are matching, which is why it's best to use macros to control the gate before you duplicate it in the first step of the process. You can also automate these parameters, allowing smooth transitions between different effects.
Here's one more example of what is possible; in this one, I have a stack of effects on both signals, and I'm automating the threshold around:
How does this work?
So this is awesome, but what is going on here? How are these gates splitting the signal up? To understand this, we need to wrap our minds around a fundamental audio concept known as "phase cancellation."
Phase in itself is simply the starting position of a periodic waveform. In this example, I have four sine waves that are identical in pitch in volume - the only thing that is different about these waves is their starting position. This is altering the phase.
In phase vs. out of phase
So if we take a single sine wave and duplicate it without changing its phase, volume, or pitch, the result will be an identical signal that is twice as loud. This is expected because the two signals are perfectly in phase, but interestingly, if you take the duplicated signal and change its phase to be exactly the opposite of the original sine wave, the result will be silence because the two signals are perfectly out of phase. This is phase cancellation, and it looks like this:
How can we use phase cancellation?
When we click the two buttons on top of our utility, we are inverting the phases of the left and right channels, resulting in the same effect you see above. If you take any channel and put it in parallel with the inverted version of that channel, you will get silence. In our case, we put the gate right before the phase inversion and then put it in parallel with the dry signal. The result is that everything but the gated part of the signal is canceled, allowing us to split the quietest part from the loudest part!
This phenomenon is at the heart of so many audio processes. It's used in EQs, filters, frequency splitters, compressors, and in our case, amplitude splitting. You can use it to create all sorts of specific effects and even increase the number of bands in the amp splitter. To do this, you would add a "middle" chain and phase cancel out the loud and quiet signals from that middle band in the same way. You would need to use different thresholds on the quiet and loud channels, but everything else functions the same.
Hopefully, this gives you some new ideas for your productions! Once you start experimenting with amplitude splitting, you'll develop an ear for which part of the sound would benefit from being processed in different ways. Have fun experimenting and bringing a fresh perspective to your effects chains!