GRANULAR AND ALGORITHMIC SYNTHESIS OF SOUND
Shana Negin and Judy Franklin
In the world of sound processing, granular and algorithmic synthesis are two methods that have been getting a great deal of attention in the past few years. In the summer of 2000 we sought to understand these methods and then experiment with combining them.
Granular synthesis employs a method of examining sound, not in pieces defined by the beginning and end of a musical phrase or note, but instead in extremely short, regular durations. These short durations are called grains.
We use two well-known computer science algorithms to explore and to
demonstrate: the bubble sort algorithm and the binary search algorithm.
The locations of the values in the list that the algorithms maniuplate are
used to choose the frequencies of a series of grains.
The result is that one can
hear the bubble sort taking place
or hear the binary search taking place. (This one loops since it finds the value in only 9 steps).
Any sound recording can be input digitally into a computer. Using sound processing software, one can access very short snapshots of that soundfile, and organize those snapshots to produce a desired (or a surprising) result. We have also used these algorithms on pre-existing sound files such as musical instrument recordings to obtain exciting and suprising results. We are in the process of composing songs incorporating them. Furthermore, Prof. Franklin has done research in the past using neural networks to make computers learn to improvise jazz music, and an eventual goal might be to continue this research, integrating the two areas.
The software we use is called (RT)CMIX, a classic sound processing system. CMIX has a unique feature of allowing the user to design filters and instruments on a programming level, instead of restricting her to a user interface. This is important to computer scientists who design algorithms and need to program them directly.
In studying these methods of sound synthesis, there are two directions that one can tend toward.
One is furthering the understanding of algorithms, using sound as a tool. The other, somewhat inversely, is using algorithms as a tool for creating music.
(Supported by the Shultz Foundation)