The identification and modelling of a percussion ‘language,’ and the Emergence of Musical Concepts in a machine-learning experimental set-up

In experimental research into percussion ‘languages’, an interactive computer system, the Bol Processor, has been developed by the authors to analyse the performances of expert musicians and generate its own musical items that were assessed for quality and accuracy by the informants. The problem of transferring knowledge from a human expert to a machine in this context is the focus of this paper. A prototypical grammatical inferencer named QAVAID (Question Answer Validated Analytical Inference Device, an acronym also meaning ‘grammar’ in Arabic/Urdu) is described and its operation in a real experimental situation is demonstrated. The paper concludes on the nature of the knowledge acquired and the scope and limitations of a cognitive-computational approach to music. Bernard Bel is an electronics and computer engineer. A founder member of the International Society for Traditional Arts Research (ISTAR) he has for many years collaborated with ethnomusicologists and musicians on projects aimed at a scientific study of North Indian melodic and rhythmic systems. In 1981 he designed and constructed an accurate melodic movement analyser, and he subsequently developed software for the analysis of raga intonation. He went on to develop software for rhythmic analysis/synthesis in collaboration with Jim Kippen. Bernard Bel is now a member of the Groupe Représentation et Traitement des Connaissances, an AI laboratory at the CNRS, Marseille.