I’m a big “music” fan. I put music in quotes as what I consider music a lot of people consider… noise. My tastes don’t really tend toward any particular area – a Venn diagram of my music collection would be colourful to say the least. Sometimes I like rock, sometimes ambient, and usually most things in between. If I’m in the mood for ambience/electronica then I like to listen to work by the likes of Brian Eno, Biosphere, Bass Communion…
A while back I tried out one of the “life simulation” games – Creatures 3. I never did get the hang of it… my attempts to create “flocks of Norns” usually involved me leaving my PC on all day so that the Norns would work it out for themselves, and coming home to a screen containing a single dead Norn. But something that struck me was the music (see, I was going somewhere) – it was rather ambient. Rather lovely. It’s only now that I realise it was generative music created by Peter Chilvers, another artist whose work I like (especially when accompanied by the vocals of Tim Bowness).
Peter has just set up a website called, oddly enough, GenerativeMusic.com, and it features a player (basically 4 flash players with different-length playlists, which works in a similar way to Brian Eno’s earlier work with tapes). It’s worth checking out, as the idea is one I’ve even toyed with for creating ambient demoworks (single progressive effect with 3-channels of sound, each channel a different length). I’ll probably never get around to doing anything with it, but if you like ambient… well, why not, eh?
3 Comments
Good stuff. This must surely be one of the last remaining untapped techniques for 4k/64k intros and the like – they algorithmically generate models, textures, animation steps, samples and even occasionally bits of text, but as far as I know everyone still stores melodies as a raw sequence of notes. One thing I really want to experiment with some time is probabilistic music – have a fixed chord sequence, and add a melody that’s random but weighted towards the notes and rhythms that are musically ‘acceptable’ – for example, sharps and flats would have less chance of occurring than natural notes. Mmm.
Yeah, Gas, it could be really interesting. Automatic music generators. I remember when I was rewriting Speccy program listings from magazines in my childhood and found such a beep “generator”, too. Well, it was just randomizing beeps but one could change the tempo! :)
For me the tricky bit is understanding music theory. It would be simple to generate music following a set of rules, for example the different-length tape system used by Brian Eno. But it requires knowledge of how music works to make it sound… nice.
An idea for demos would be to base the sound on the source material of the generated effects. Assuming that the effects themselves are based on tables of data, then the sound should in theory match, or follow, the effect as it progresses.