StuRants

Music of the proteins

Late last year I briefly entertained the idea of putting together some sort of program that sang the numbers in the sequence of pi at one note per second for one year. My first step was to generate pi to 20 million decimal places. (I didn't do this myself of course, I got a microcomputer to do the deed.) The next problem was working out how I'd do it, and here's where I got snagged.

To determine the pitch of the notes was easy enough - I'd just need to convert the numbers into base 12. But then, how to calculate the octave. Uh, maybe flip the number into base 8 and use that for the octave. Okay, so what about tempo and volume changes? Well... And the voice is going to be dull on its own - we really should score it for multiple instruments. How do we get pi to conjure that up? Ah. Hmm.

As you can see, the idea quickly became quite strained, with pi to be required to chopped up in so many ways that using pi as a generator would be meaningless - it'd be the arbitrary decisions used in crunching the numbers that would be doing most of the work in the piece. So basically my pi plan never got cooked. (Ho. Ho.)

Then last week I read about a project to make music from amino acid sequences in human proteins. Rather than converting an acid into a note, they've made them play chords, so that the resulting music is more harmonically interesting.

Apparently one of the applications for this approach is to be able to aurally determine familiar chord patterns to see what's going on in these proteins. If that is useful, then great, but otherwise the proteins make for pretty poor music. It's just plink plonk plonk plink all the way down the line.

Sure, the protein pianists can refine their algorithms and get more subtlety in there, but really, we should take a step back and ask why do some people (not many, but a fair few academic types) make music out of number irrational numbers, or fractals, or generative numbers, or scoring symphonies via the I Ching? The answer, I think, is simply to see what it sounds like. Fair enough. But this doesn't mean, having gone to all this effort, that the result is must be good music. In fact the result is barely music at all - I'd rather hear a chimp at the piano than a protein (much rather, in fact). It's a situation, I think, where the process of deriving the music has hijacked the value of the outcome. It may be that some day some computer will pass a sort of musical Turing test (and what would that achieve other than the redundancy of advertising jingle writers?), but the programming involved will be monumentally difficult. It all seems a bit pointless to me.

Lest this seem like yet another post where I rag on avant garde music, I am very much for things like the 2003 attempt (well, do we know it was successful?) to recreate the sound of the Big Bang. (Admittedly this has nothing to do with avant garde music, but it does mean I end this post on a positive note.) Recreating sounds of things that we cannot hear using accurate enough measurements (those recordings of the atmosphere of Titan using a sampling frequency of 1Hz didn't really do it for me) satisfy a natural human desire to experience aspects of the world we can't otherwise. It would be a great outreach tool, too.

And I'd love to hear more Radio Astronomy recordings, preferably with a CD-quality dynamic range (ie 20Hz to 20kHz)*. Thanks!

* From what little I know of this subject, the bandwidth normally sampled is considerably less than 20kHz. In fact 20kHz is probably considered grotesquely wasteful.

Posted by stuart at May 5, 2007 5:02 PM