4 posts • joined Thursday 4th September 2008 11:49 GMT
Nice one Guardian.
Funnily enough I 'ad that Bertrand Russell in the back of my cab once. "So, Lord Russell," I asked 'im, "what's it all about then?". And do you know what, 'e couldn't tell me! Mind 'ow you go gov...
There's a good reason it sounds rubbish: it's a produce of e-Science, the paranormal belief that if governments give a stupidly large portion of their research budget to computer scientists the world's problems will be overcome because the rest of science is trivial, all that's left to do is crunch the numbers quickly.
Accurately modeling sound generation and radiation in musical instruments is really, really hard. The things that make real instruments sound different from casio presets are subtle effects such as all sorts of small but significant nonlinearities, polarity mode coupling, dispersion etc. We don't completely understand how these work together even in relatively simple plucked instruments. We could sidestep the problem of having to understand by doing Direct Numerical Simulation of the underlying equations (Navier for the vibrating parts, Navier-Stokes for the air). Two problems there: 1) the resulting problem would be so huge that the e-scientists precious grid wouldn't stand a ghost of a chance and, more importantly 2) we don't yet know well enough how to model complicated things like wood (let alone varnish) and we're only just starting to understand plucking, so we wouldn't know what parameters to give to the program.
Checking the astraproject website I see that In true e-science fashion all of these issues have been handily ignored and instead the team behind this 'research' used Julius O Smith III's vastly overhyped digital waveguide synthesis approach, which is popular precisely because it throws out everything that makes instrument modeling hard so as to run quickly. Unfortunately the things that make the modeling hard are what make the sound interesting and the results are there for you to hear. Digital waveguides are a necessary evil if you've got to model the instrument in real time so that the performer can play with parameters while a note is sounding and hear the effects as she does. But if you had to model any wavebearing structure (like a car or a bridge, say) and cared at all about accuracy this method would be at the very back of the queue after pretty well every other numerical technique.
Tony Hey (Microsoft VP who used to be in charge of diverting UK taxpayer's money to this sort of stuff) should be forced to listen to no other music than this for a year to remind of what he's done.
- Xmas Round-up Ten top tech toys to interface with a techie’s Christmas stocking
- Xmas Round-up Ghosts of Christmas Past: Ten tech treats from yesteryear
- Review Hey Linux newbie: If you've never had a taste, try perfect Petra ... mmm, smells like Mint 16
- Analysis Microsoft's licence riddles give Linux and pals a free ride to virtual domination
- I KNOW how to SAVE Microsoft. Give Windows 8 away for FREE – analyst