Structure and expressive timing in music performance

The research presented is based on the study of music, the study of mind and the study of machine. Most of it deals with the low level rhythm perception mechanisms that process the perceived time intervals roughly corresponding to a note, and infer rhythmic categories, a sense of tempo and local deviations thereof, and expectations for the events to come. This main thread of the work consists of the first four articles. They deal with different methods of quantization, propose a connectionist model and compare it to a symbolic one, and elaborate an expectancy measure that can be used as the basis for higher level rhythm perception processes. Thus starting from the perspective of a technical tool the work develops towards cognitive theories of human rhythm perception. Then two papers follow a side-track that deals with AI programming in LISP. They explain the programming style used in the various articles and apply it in a case study to a large computational model. The next three articles follow a main line that studies the expressive timing signal extracted by quantization. The first of these tries a statistical approach, analysing the tempo variations in a performance using autocorrelation. The second, in the form of a fictitious story, warns against a simplistic notion of tempo curves by showing that any transformation or manipulation based on the implied characteristics of such a notion is doomed to fail. The third concludes the thesis, linking expressive timing and structure in an attempt to enable transformations of expression. The motivation for this last article is the need for measurement instruments and tools that can cope with the complexity of performance data and are much more sophisticated than tempo curves. It assumes an intimate link between expression and structure - or rather the foundation of the concept of expression on structural musical units