this thing took ages to render. basically i converted the whole track into a mono file and then into an array of floating point numbers. from there, a Haar wavelet transform was applied that andy, my professor and i have been working on for quite some time. once the wavelet coefficients were derived and compiled into an array, i chunked it into several 512-float subsets, and randomly scrambled their order. then i flattened the array back to a single dimension. finally the inverse function resulted in a string of floats from -1 to 1, which were written back into a buffer and saved.
the Haar wavelet is the simplest wavelet invented, and the first discovered. it is the first in the Daubechies series. we coded this transform in supercollider from an example we found in Java. the alterations to the coefficients are the latest in my search for meaningful manipulations. i figured it was only right that i use a new track by andy as the source material, because he and i toiled long and hard to figure out what exactly my prof was talking about in our quest for the inverse function.
the cool thing about this guy is the constant transpositions of somewhat recognizable material. another good one was the stutter effect, which i will post after this. this piece of music is possibly the most gratifying thing i've posted in a really long time. this may not sound like much, but the supercollider program was hung for a solid hour trying to render the soundfile array from the coefficient data.
i'm so thrilled to have gotten any sound out of it at all, much less this stuff, which i absolutely adore.
