Algorithms are sets of rules that produce a finished state. Using algorithms to make music has a long history. As Jon Appleton observed: In both the Middle Ages and the twentieth century, composers have celebrated music's link with the logic of mathematics by introducing parametric systems of organisation (primarily in the pitch domain), which are largely unrelated to aural perception. In the Middle Ages these techniques were invariably hidden, existing below a surface that conformed to stylistic norms. In the twentieth century, some composers used a technique that introduced a novel way of ordering pitches, but did so within the context of traditional musical forms. (Appleton, J. (1992) Machine Songs III: Music in the Service of Science - Science in the Service of Music, Computer Music Journal, vol. 16, no. 3, p. 18). Algorithmic music has expanded to include stochastic and chaotic, formal and linguistic, generative and quasi-neurological algorithms. Artificial Intelligence, in all its various forms, provides a fruitful area for much digital music creation today. The purpose of this project is not necessarily to create new algorithms or to investigate this field, but rather to understand the relationship between the musician, the computer and the algorithm. The popular puzzle Sudoku provides a useful tool for this purpose. Sudoku may be found in many newspapers, puzzle books or on the web. The rules are simple: fill the 9 • 9 grid so that each column, each row, and each of the nine 3 • 3 boxes contains the digits from 1 to 9
Make three digital musical versions of a Sudoku puzzle: 1 computer-controlled; 2 'user'-controlled (where the user is someone other than the person making this project); 3 composer/performer-controlled. The numbers may map to any parameter of the music (sounds, rhythms, intensities, etc.)
Mapping decisions are significant here. The computer-controlled version is straightforward to execute, but how can the results be made musically interesting? Giving the user control may require some kind of training, if they are to complete the puzzle successfully. Is it the solution, the original, or the processes involved in solving the puzzle, that is being mapped? If this is to be performed by the composer, to what extent (if at all) can the algorithm be used to reflect compositional intentions?