Ten Minute Master: Generative Music
If you need inspiration or simply want your music to take a completely new direction, try a walk on the generative side. Rob Boffard is your guide. What if you could get a computer program to produce and play your music for you? You’ve got to admit it sounds like an odd idea – the […]
If you need inspiration or simply want your music to take a completely new direction, try a walk on the generative side. Rob Boffard is your guide.
What if you could get a computer program to produce and play your music for you? You’ve got to admit it sounds like an odd idea – the concept of DAWs and plug-ins taken way too far, something that removes the human operator entirely and replaces it with an AI, Simon Cowell’s wet dream. But not only do these programs exist, they form the backbone of a legitimate type of music – one which, as we’ll show, can be extremely useful for any musician.
It’s called generative music, and the software that produces it quite literally creates its own music, with only minimal input from a producer. So why bother? And how does it even work, anyway?
It’s not quite as bad as you might think. For starters, a computer program being able to self-produce a track that could chart on the radio is a long, long way off. So while we might be convinced that someone like Skrillex is actually a robot, we’re not entering an age of redundant producers just yet.
Rather, generative music relies on a program governed by a series of rules set by a human producer. The producer picks the number of voices, ranges in pitch and timing, and lets the program do its thing. What pops out will – all being well – be something musical (and even listenable).
There’s an analogy for this which we can ’t improve on (we haven’t been able to find out who first thought of it, but it works beautifully). Imagine a lane in a bowling alley. Instead of gutters, you have raised walls, so that any ball that hits them will keep rolling down the lane.
Now, imagine releasing several bowling balls into the lane. They’ll roll around and bounce off each other until they reach the end. But collect them up and do it again and they’ll collide in different ways, following different paths down the lane. The nature of the movement is the same, but the actual movement patterns aren’t.
The music (the ways in which the bowling balls collide) is constrained by the rules you set (the lane, the walls on the sides). You have little control over the exact paths of the balls, even though you maintain control over their environment. What this means in music terms is an ever-changing, constantly surprising piece of music.
The point of doing that? Randomness. By creating something and letting it run free, you’ll discover patterns, ideas and motifs that you might never have thought of. Generative music is an incredible tool for inspiration. Since a plug-in lacks tastes, prejudices and distractions, it’s the ultimate tool for generating randomness. And, of course, there are plenty of artists – think Brian Eno and Autechre – who are big advocates of using generative software in recording. With a little experimentation, you’ll find that the software can take your music and your ideas in completely unexpected directions.
Break it down
So what does a piece of generative software look like? The good news is that you don’t have to possess any special skills or be a whiz with coding to use them. Most will slot into your DAW as plug-ins or work as standalone programs. And most are very easily understood.
The most intuitive we’ve found is CEMA Research’s Nodal. As a way of demonstrating generative music, it takes some beating. You’re presented with a grid, and on that grid you can place various little nodes – points at which the music will do something. You can assign different voices or sounds to different nodes and link them together via paths (or ‘edges’, as Nodal terms them). It’s these paths that make things interesting.
A node can have more than one path branching off it, and you can tell Nodal to follow certain paths at random, at certain times or when a certain note/voice is triggered. You can even tell it to start adding in parameters whenever the music travels along a certain path.
What this means is that by using the nodes and paths to set the rules, you can tell the music what conditions to follow. But once you press play, the results are always a little unexpected, and what starts off as a simple note can, once it follows the paths you set for it, morph into a lush and complex piece of music.
Most generative software follows a similar idea. You set the number of voices and the parameters, then press play and let things roll. There’s plenty of excellent software available; besides Nodal, we particularly like Intermorphic’s Noatikl plug-in. This one takes a slightly more functional approach to things, but allows a more direct integration of MIDI data into the proceedings. Noatikl’s predecessor, Koan, was used by Brian Eno to create a project he called Generative Music 1.
One thing you might encounter when you first start tweaking generative software is frustration. We’re so used to having our music software perform in predictable ways according to the inputs from our controls that it’s often quite difficult to wrap our heads around something which removes some of that control. And trying to incorporate a generative piece into an existing drum break (for example) will probably have you hurling your computer through the nearest window.
You might expect us to advise you to take it slow and to start small. We won’t. We find that the best way to approach any generative software is to not worry about structure. Add loads of voices, draw bizarre patterns, program in rules at random. Generative music thrives when you have fun and experiment with it, and while psychologically it might take a little adjustment, you’ll need to start going crazy if you want to be truly inspired.
The one thing you’ll find during any foray into generative music is that it can turn into something very dry and academic very quickly. For some reason, it’s the kind of concept that attracts considered study. We don’t mean to suggest that we’re against academic interest in a form of music, but it does tend to shroud it in a fog of impenetrable terms.
That being said, there are a few theories of generative music worth knowing about. Think of them as ways of viewing the genre and approaches to working in it. First put forward by René Wooller at Sydney University, they’re a good way to understand the different forms of generative music.
The first and most intuitive theory is called creative or procedural theory. Essentially, this is when the producer intervenes by setting the rules that govern how the program generates its sounds. This is the most common form of generative music, the easiest to get involved in, and the easiest to understand.
Then there’s biological theory. This describes music that cannot be repeated. With most generative software, the music will follow a pattern – at least at first. But when things get biological, there’s no pattern, and any sound that gets played is generated entirely at random. Wooller, in another rather nifty analogy, uses wind chimes as an example.
There are further theories, but this is where it starts to get a little more abstract and a little less useful. There’s behavioural or interactive music, which relies on having the program duplicate certain behaviours even as the sound changes; and linguistic music, which is a lot more rigid and predictable.
The good news is that you don’t have to use or worry about these theories too much to get the best out of generative music. All you need is a little curiosity and a willingness to go down some rather strange paths.