An efficient algorithm for motivic pattern extraction based on a cognitive modeling
Résumé
This paper describes a computational model for discovering repeated patterns in symbolic representations of music. Patterns are discovered through an incremental adaptive identification along a multi-dimensional parametrical space. The difficulties of pattern discovery mainly come from combinatorial redundancies, that the human cognitive system is able to control efficiently. Our research attempts to reconstruct these principles. A specificity relation is defined between pattern descriptions, unifying suffix relation-between patterns-and inclusion relationbetween multi-parametric pattern descriptions-and enabling a filtering of redundant descriptions. Successive repetitions of patterns imply another kind of combinatorial proliferation, which can be managed with cyclic patterns. By reconstructing these redundancy control mechanisms, musicology-relevant analyses can be automated. Current researches include the enrichment of the model, which was primarily dedicated to the analysis of monodies, with polyphony management mechanisms. The system may be used either for musicology researches, as an improvement of traditional analysis techniques, or for industrial application to automated analysis of musical databases.
Origine : Fichiers produits par l'(les) auteur(s)