On Algorithmic Representation of Music Style David Cope Professor of Music, Univ. of California, Santa Cruz
Overview EMI (Experiments in Music Intelligence): represent music styles based on linguistic principles Language Parsing Augmented Transition Networks (ATN) Non-linear composition Future work
Language Parsing Emphasis: hierarchy (most to least important) and abstraction of function Ex. I am driving to the country. I am driving the country. I am driving to the ( ). I country.
Language Parsing (cont.) Hierarchical Parsing (Arrow: logical modifier processions) Abstraction of function
Musical Parallels Schenkerian layer analysis: prolongation of tonic from the first to the last C Hierarchical Parsing
Music Parallels (cont.) Abstraction of function EMI Symbology: OSAC (identifiers, priority order) O: ornamental S: statement A: antecedent C: consequent Level of importance: 1-3 (least to most significant )
Augmented Transition Networks (ATN) Allow finer detail between context-free and context driven environments
ATN (cont.) Music example
Non-linear Composition Top-down method Music has an end. Mid-level communication No backtracking
Example A Back-like keyboard invention
Future Works Refine ATN engine More accurate representations of style integrity