Wired SEPTEMBER 28, 2012 - by Ian Steadman


When you think about it, recorded music is something of a historical oddity. Imagine telling a visitor from 1650 that we like to listen to the exact same performances, over and over again, without any variation - they might well be baffled. Or, as Brian Eno once said, perhaps it is our grandchildren who will be the baffled ones.

Scape, a musical app developed by Eno alongside composer and programmer Peter Chilvers, is a kind of album that will never sound the same twice. It's a continuation of Eno's work with generative music, compositions which change every time they're played as the systems behind them introduce some kind of randomness. Starting with an empty screen, shapes can be arranged to create literal "soundscapes" where they each represent a different sound. Their positions relative to each other dictate their behaviour.

Eno, one of popular music's most important and influential figures, has used generative techniques to compose music for many years, popularising the term with the album Generative Music 1 in 1996. Chilvers and Eno first worked together on the soundtrack to Spore, where the generative music matched the sandbox evolution of the gameplay. They then worked together on apps called Trope and Bloom, both more primitive versions of the "soundscape" idea behind Scape.

We're used to the ideas of live music versus recorded music, and the different expectations they bring, but generative music introduces the unpredictability of live playing to machines. The description of Scape as "music that thinks for itself" imbues computers with a kind of creative responsibility, which might seem a contentious notion. Can an algorithm be creative in a way that we would recognise?

Wired met with Eno and Chilvers at Eno's studio in Notting Hill to discuss the development of Scape, and what it means for the future of generative music as a popular art form.

Wired: You two have worked together before on similar apps, but this seems a much more elaborate setup - to what extent is this a progression of those same ideas?

Eno: They're progressions from Spore, where we wanted to make music that wasn't repetitive or based on simple loops going around, but was based on the idea that any scene could have a sonic character - but it wouldn't be exactly the same each time you visited. So we came up with this idea of loading the game with what we called "shufflers" - now they're called "elements". There were simple rules for how those things combined and played out - this thing only happens when these other two things are happening - and so on. This worked quite well in Spore, and Bloom was one of the spinoffs from that. Bloom used one sound that we used in Spore. I think it was a big step forward, actually, that you could create a piece of art and not just an interesting technical curiosity. Scape is another step, to say, "what if we used a bigger family of elements?"

So where do you come up with those different elements, and how they interact with each other?

Chilvers: A lot of the sounds came out of a mixture of existing pieces by Brian, so there was already an existing combination that would sit well together. I think it's partly down to the nature of the kind of music that Brian has been creating that they actually sit well with most sounds in this universe.

Eno: That's right, it's a kind of universe of sounds that work well together, and they can have quite ambiguous tonal relations with each other. You couldn't imagine doing this if you took apart, say, a Clash album, and you took apart the drum parts, and the guitar parts, and the bass parts, and let them float freely and recombine. It probably wouldn't get you a very interesting result. But this music was made from the beginning in that way. It was made on the idea that the elements within it were not in a fixed relationship to each other, they didn't have to be in just one relationship. The discovery that we made was that you could take three or four of those elements from one piece and three or four from another piece and - ah! - they can work together. It's been treated like a composition process from the beginning really.

I've seen you talk about it generating its own compositions - that it has its own creativity. Do you think that a machine can generate real music without human input that we would still value?

Eno: It's not without human input.

Chilvers: It's not real-time human input, I suppose. To come back to a metaphor that Brian often uses, it's more like gardening; we're providing a set of seeds more than anything else, but those seeds were already designed elsewhere and they were already given that initial kickstart of creativity there. It's a very unusual type of creativity, it's quite open-ended.

The interface adds to that. I don't want to call it messy, but it's not rigid. You can place things wherever you want. That's very deliberate I imagine?

Chilvers: In fact it's something I found slightly alien. We actually bent the rules slightly so you can place things just off the screen, so they're just spilling out over the edge.

Eno: It also gives you a different way into the composing process because you can compose a piece just by making a picture that you like, which I find quite interesting.

Chilvers: There's an interesting example of that actually. If I just load the very first one, "Icon", the name isn't a coincidence. We have a graphic designer, and he wanted some sort of suggestion of what the icon for the app would look like, so I mocked something up with Scape. All I was trying to do was get one of each type of element, and by the end of it we thought "that's a really nice piece". So it ended up becoming the opening. So this is a great example of something with no real sense of what the sound will be, just trying to do it naturally.

It's almost like a twenty-first century version of Oramics isn't it?

Both: [Laughter] Yes, yes.

Brian once said that in the future we'll have three kinds of music - live, recorded and generative - and that our grandkids will look back and say "why were you just listening to the same thing over and over?" Do you see generative music becoming more common?

Eno: Well, in a concert of course, you're in a situation where you won't get identical repetition anyway. Well, actually, that's no longer true any more, since bands now use samplers and sequencers. But in general the idea of going to a concert is that you expect to see something unique to that evening. It's a clever con [laughs].

Do you think we'll see generative music becoming more common?

Eno: I think we are seeing it now, you know. Bloom was a very successful app, a lot of people bought it and a lot of people love it, and still use it. I remember about a year ago, I phoned up a graphic design office and I heard Bloom playing in the background! I asked the receptionist what the music was they were playing, and she said "oh I don't know, they're always playing it, it's nice, isn't it?" So there was the example of generative music in action.

How about the potential for automated music heralding a shift in the way we value music? I remember having arguments in the school playground with friends over real instruments versus synths, but part of that is also a sense of the importance of intent and authenticity. Where do those ideas fit into music like this?

Eno: Well, you'll note that we haven't tried to make a flavourless tool that can do anything. It makes the kind of music we like, so there's quite a lot of intent in this, and there have already been a lot of aesthetic choices made. When somebody gets this and makes a piece of music with it, they're making a piece of music in collaboration with us. It's not going to suddenly burst into afrobeat or anything like that.

Do you see that being something that you'd like to expand? Would you want to open the coding behind this so fans could make an afrobeat version, for instance?

Chilvers: It's one thing we've contemplated, because it would be possible. Partly as a programmer, that doesn't feel particularly interesting to me. It's far more enjoyable doing something that's very defined. I think what's been nice with Bloom and Scape is that you need no musical experience, no skill to operate them. I think you need taste but not skill, and I think to some degree it teaches a kind of taste. It's been good at teaching people to slow down a bit and use less notes.

Eno: I think when you make something you are always offering some choices and denying others. That's in the nature of anything that you make as an artist - you're trying to make a thing, not everything. I remember I had done some artwork a long time ago, I had this big piece that showed on a plasma screen, and I was very pleased - and then somebody looked at it and said "it's quite good but the screen stays still". I said, "what do you mean?" She said, "well I think it would be much better if the plasma screen rotated". Jesus! Don't want a lot do you? [Laughs] You just accept there are certain things you're not going to do in this piece, and part of the process of making this was deciding not only what to include but what to leave out. We could have put lots of different options - shall we have tone control? Shall we have mood controls? What you're always dealing with is this playoff between options and raport. Why is it that guitar players are still making interesting music on this primitive, ridiculous instrument? Six strings, arbitrary length, yet they still do more interesting things in general than synthesizer players do, and I think it's because they're not baffled by options. They quite quickly know what the instrument can do, and they start doing something with it - whereas my experience of synthesizer players is that they're constantly looking through manuals and not actually playing! [Laughs]

So in that case, if there was to be a Scape 2, there wouldn't be more options, but different options?

Chilvers: Yes, I'd say so.

Eno: Well I think one of the things we might well do once, and if Scape becomes sort of settled and established, is to release new libraries of sounds that can be used but that would be on the basis that people have already understood it.

When they're ready to take the training wheels off?

Chilvers: Yes. And I think with these things you have to ask yourself "is this in the way?" For example, you can't name your own Scape, because, if, as a photographer, I had to stop every time and enter a name for every photo I took, I wouldn't use my camera very much. I like that it's very immediate and not fussy. But one of the things that's fascinating when you develop an app is that it's quite a two-way process with the buyers, almost every comment you get is "I love this, this is great, but it needs to do this". With Bloom I think people wanted more sounds, but that was partly down to a kind of misunderstanding of what Bloom was. We wanted Bloom to be a single piece. At one point we tested out about fourteen new sounds and in the end whittled it down to just one extra one, which is a very subtle variation on the original. Which I think did improve Bloom, but it did still feel like Bloom.

How about the evolution of these sounds? Do you want people to share the sounds they generate in Scape, and then they can download it and take it from there?

Eno: This is another thing that we thought about quite a lot. Peter put this very well, saying "I'm sure a lot of people would want to share them, but I don't know how many people would want to hear them". It's very nice, the thought of sending someone something, though.

Chilvers: I like the idea of emailing someone a Scape song and them changing it, because as soon as you get a Scape from someone else it doesn't go into your library, it goes into your "create area", and you can meddle with it. You aren't obliged to save it, and you can change it, make it your own, and email it on.

It reminds me of an experiment at Imperial College London where they took four random loops of noise and got them to "breed", and then bred the offspring to create new sounds, and so on, to create music. This seems a more structured and artistic version?

Chilvers: Yes. Technically it's a finite space of sounds you can make with Scape, but it's a pretty huge one. There are something like ninety different elements, and there effectively are ninety different programs behind them. They will be thinking, and some of them will specifically respond to different things, like "three other things have made a sound so I won't for a while", or "everything's noisy so that's definitely a time where I want some attention".

Eno: They're not only sounds, they're personalities as well, in the sense that they have a relationship to each other. People might never notice this thing that only behaves that way because it's in company with that. If it isn't in company with it, it will behave differently. I think people will get a sense of this after a while.

That makes it feel like an instrument rather than just a synthesizer or a sampling program.

Eno: That's the main difference between normal instruments and synths. Normal instruments are made up of thousands and thousands of molecules, whereas electronic instruments are the amplified movements of a few atoms. So they're much more deterministic than normal instruments.

Chilvers: There's something of a lack of eccentricity to a lot of electronic instruments, I feel. That's the thing I enjoy the most about Scape, is making things a bit chaotic and unpredictable. Fuzzy, for want of a better word.

So, what next after this?

Eno: We've got a few ideas.

Chilvers: Nothing we can talk about at the moment, but we've really only just scratched the surface yet. I think what's been interesting has been seeing Scape become part of a much wider spectrum that's including certain games. I keep coming back to Osmos, which I find fascinating. It's as close as you can get to Bloom, but as a game. You're starting to see more and more of these on the iPad, things that are quite tactile and interactive, and it's just simply that the experience of playing is enjoyable in itself. You don't necessarily care about scoring, or achieving things, or beating your friends. It's just really rather pleasant to be involved in it.