Towards a Generative Theory of Interpretation for Performance Modeling

Abstract

Performance structure is described in terms of segmentation and classification of segments. Types of segments are associated with specific execution techniques. Similarly to speech, segments are shown by “performance waves” in timing and dynamics with a single common peak. Performance waves are arranged in a hierarchy corresponding to the form of the piece which can be understood rather subjectively. The interpretation score may look as an orchestral arrangement, where segments are classified with respect to their type and/or execution techniques (as if played by different instruments). We apply our approach to theatrical interpretation as well. Each character is described by means of several “basic types” (love, hate, sincerity, hypocrisy, etc.) associated with specified intonation and gesture. Then the role is split into segments, each being associated with one of the basic types (cf. with switching from one execution technique to another in a musical performance). It is a further development of the interpretation technique in the theater of Ancient Greece where the actors changed masks (of anger, joy, melancholy, etc.) to show emotions. The approach is intended for developing music software as well as for music and theatrical education.