So you basically have s=f(t) where s is the output sample, t the sample id and f a simple expression as the one I put on the title.
The thing is you can actually build quite complex song with it !
This technique has been called bytebeat and the main article describing it was made by someone named "viznut".
See the original article here http://countercomplex.blogspot.fr/2011/10/algorithmic-symphonies-from-one-line-of.html or here http://canonical.org/~kragen/bytebeat/
and and example with trippy visualizations :
I surely will include one or two of them in a test audio program on the sdk .. In the meantime here is the code :