Evaluating a Swarm Algorithm as an Improvising Accompanist: an Experiment in Using Transformed Analysis of George E. Lewis's 'Voyager'

How can one creatively and analytically approach the design and implementation of an improvising computer accompanist? Given such a computer accompanist, how does one evaluate it? I made a speculative high-level mapping of components and functionalities in George E. Lewis's Voyager (1987) according to published information about the work. [4] Transformations of this analysis resulted in an algorithm that I could implement in an interactive improvising computer accompanist which I call Favola. I tested this computer accompanist with three instrumentalist improvisers, two of which were able to produce convincing music with the system. The use of analysis and transformation of this analysis has fuelled the creative and technical development of an arguably successful expert system. I consider directions for potential robust evaluation of the 'convincingness' of the improvised music resulting from the system