Last week I've been playing around a bit with sound synthesis, as I was interesting in seeing (or rather hearing) what it takes to get a decently human voice sound from scratch. Given the horribly low amount of documentation you can find on sound (compared to graphics), of course, it was mostly a matter of trial and error, but truly an amazing learning experience. Working purely from code, building up all the harmonics waves manually, was a nice addition to the challenge as well.