Dot Matrix was an exploration into pattern-based generation algorithms for melodies and stochastic models for rhythm and accent structure. The algorithms for this piece were initially developed to model classical music and jazz solos, but also turned out to be well-suited to electronic dance music when the pattern length is more strictly constrained. Each melodic pattern is a short sequence of 2-4 numbers, which must be instantiated as a melodic contour from a particular starting pitch within a scale. Pattern instances are then combined together in sequence with a path-finding algorithm that tries to find a smooth connectivity between them. Harmonies are created using delay lines to repeat a sequence over itself. Rhythms and accent structure for the bass and drum kit were generated using modified versions of the metrical grid model from Generative Theory of Tonal Music and David Temperley’s work – the visual representation for which is a collection of stacked dots over each beat. Another algorithmic composition system, Kulitta, features in this system as a “guest performer” on hand drums. The dancing dots in the visualization react to features in the music while also exhibiting attractor-based behavior in some sections.
Donya Quick is a Research Assistant Professor in Music and Computation at Stevens Institute of Technology. Her research explores the intersection of artificial intelligence and computational linguistics with music. She completed her PhD at Yale University, where the subject of her work was an automated composition system called Kulitta. In addition to continuing work on Kulitta, Donya is also involved in the MUSICA project for interactive music creation through natural language, which is part of the DAPRA Communicating with Computers program.