Abstract:

The process of composition can be seen as sequence of manipulations on the material. In algorithmic composition, such sequences are prescribed through another set of sequences which yield the algorithm. In a realtime situation the sequences may be closely linked to the temporal sequence of the unfolding musical structure, but in general they form orthogonal temporal graphs on their own. We present a framework which can be used to model these temporal graphs. The framework is composed of layers, which–-from low to high level–-provide (1) database storage and software transactional memory with selectable temporal semantics, (2) the most prominent semantics being confluent persistence, in which the temporal traces are registered and can be combined, yielding a sort of structural feedback or recursion, and finally (3) an event and expression propagation system, which, when combined with confluent persistence, provides a hook to update dependent object graphs even when they were constructed in the future. This paper presents the implementation of this framework, and outlines how it can be combined with a realtime sound synthesis system.