Distributed Computation for the Masses
Related link: http://www.thalassocracy.org/soundmosaic/
"soundmosaic is a distributed system for constructing an approximation of one audio sample out of pieces of other samples -- a kind of "photo mosaic" effect in sound. This is hackerware for the moment, so don't expect to be able to use it without getting your hands dirty.
The soundmosaic algorithm is this: Split the original sample up into even-sized chunks. For each chunk of the original sample, find the closest match in the source sample, and replace. (The difference between two chunks is defined as the area between the waveforms.) In order to eliminate volume as a factor, the chunks are normalized prior to comparison, and the selected match is adjusted to the volume of the original chunk before it is written to the output file.
In order to find matches good enough to make both input samples recognizable, we need a tremendous source sample, and a tremendous amount of data storage and processing to go with it. Distributing the system helps to handle that load so that a decent result can be achieved in a reasonable amount of time. "
...
"each of these [13 second] mp3s represents something like a total of one day of processing split between my computer at home and my laptop"
I have never seen something like this before: a completely new functionality enabled by the kind of massive computing resource formerly available only to large projects becoming available to a project with no funding whatsoever. I wonder about what kind of wildly inventive things we may see as distributed computation becomes more widely available?
Categories
Audio
Read More Entries by Lucas Gonze.
