Suzanne Kaprov is an artist who creates among a variety of unique works or art, pieces centered around...well, pieces of puzzles. Her work was deeply influenced by her friendship with John Cage.
"First I take a blank, factory-produced puzzle. After coding the backs of each piece so I can later reassemble them, I then disassemble the puzzle and place the pieces into a bowl. The pieces are mixed as I would shuffle a deck of cards. I then randomly pick a puzzle piece from the bowl, ready for painting. In another separate bowl, I place a hundred individual stickers, each one numbered and labeled with a different (written) color, for a total of one hundred color choices. I select five colors for each piece, one by one, randomly picking them from the bowl. Each sticker is always returned to the bowl, which is re-shuffled again. Once five colors are picked, the individual puzzle piece is ready for painting. I repeat this until all the pieces are done and when the puzzle is reassembled, the completed 'big picture' finally emerges."
I was asked by Ear to the Earth to use a recorded text of Suzanne speaking 100 words and create a process analogous to her paintings.* My software diced up the recited text, creating an individual audio file for each word. A series of one dozen audio effects exploiting eight channel (3-d) audio were developed. A random number generator selects how many effects are applied, in what order, the values for all of their variables, and the amount of overlap between files. Examples of the "remix" from eight channels down to stereo can be heard here.
*The process described on the E2E website is not exactly what I used.
In the fall of 2011 I created a program from scratch designed to generate pieces of "tonal" music. My initial work relied on Markov Chains, but I found the results to be rather predictable once a model was decided upon. I then decided to use probabilities with shifting weights in conjunction with random number generation to make all "decisions." The software seeded the piece with initial selections for:
Using the irrational number Phi as a measure of proportion and timekeeping, the system kept running averages of times, frequency of occurrence, and types of previous events in order to change timbres, exchange melodic fragments between voices, modulate, apply frequency and amplitude modulation, and change global as well individual dynamics.
The timbres themselves were generated with a separate wavetable for every pitch executed, giving the instruments the subtle appearance of register found in acoustic instruments. Random number generation was applied continuously to details throughout the process in order to increase subtle complexities, including a random scalar value to every note generated in order to humanize the execution of the dynamics.