When I started the Fnk project, one of the many things I was anxious to play with is audio analysis. The main reason being the fact that I know too little about it. So here goes a preliminary post on the subject; no conclusions, but a few good links for future reference.

For the kind of visual work I’m aiming Fnk to deliver, audio analysis is usually done with what’s called a Fast Fourier Transform (FFT). I’ve worked with FFT in the past – my Rorschachoscope project uses certain FFT values to detect when a person claps his hands (in order to save its current image to a file) using vvvv – but that’s where my knowledge of the subject stops.

When prototyping multimedia installations, though, beat detection is usually what one’s interested in, and that’s where FFT comes in. Robert Hodgin has some interesting information about that here and, more recently, here. His written words make it clear that it’s a difficult subject, but his filmed works show that there’s some great results possible nonetheless.

One other great link, this time focused in Actionscript, is this demonstration by Анатолий Зенков. It also uses FFT, and even includes its source. Also check these other links on the same subject; they’re very inspiring for me, and for the kind of work I want to make possible with Fnk.

And at last, but not least, Mr. Doob has some experiments in the same subject done with Actionscript and Papervision3D – check one of them here, and use the top menu to see the others. Mr. Doob was also the guy who sent me Анатолий Зенков’s link, so he wins two inspirational points on this post.