Ontology of Music Performance Variation
by John R. Taylor
University of Sydney
Performance variation in rhythm determines the extent that humans perceive and feel the effect of rhythmic pulsation and music in general. In many cases, these rhythmic variations can be linked to percussive performance. Such percussive performance variations are often absent in current percussive rhythmic models. The purpose of this thesis is to present an interactive computer model, called the PD-103, that simulates the micro-variations in human percussive performance. This thesis makes three main contributions to existing knowledge: firstly, by formalising a new method for modelling percussive performance; secondly, by developing a new compositional software tool called the PD-103 that models human percussive performance, and finally, by creating a portfolio of different musical styles to demonstrate the capabilities of the software. A large database of recorded samples are classified into zones based upon the vibrational characteristics of the instruments, to model timbral variation in human percussive performance. The degree of timbral variation is governed by principles of biomechanics and human percussive performance. A fuzzy logic algorithm is applied to analyse current and first-order sample selection in order to formulate an ontological description of music performance variation. Asynchrony values were extracted from recorded performances of three different performance skill levels to create “timing fingerprints" which characterise unique features to each percussionist. The PD-103 uses real performance timing data to determine asynchrony values for each synthesised note. The spectral content of the sample database forms a three-dimensional loudness/timbre space, intersecting instrumental behaviour with music composition. The re-parameterisation of the sample database, following the analysis of loudness, spectral flatness, and spectral centroid, provides an opportunity to explore the timbral variations inherent in percussion instruments, to creatively explore dimensions of timbre. The PD-103 was used to create a music portfolio exploring different rhythmic possibilities with a focus on meso-periodic rhythms common to parts of West Africa, jazz drumming, and electroacoustic music. The portfolio also includes new timbral percussive works based on spectral features and demonstrates the central aim of this thesis, which is the creation of a new compositional software tool that integrates human percussive performance and subsequently extends this model to different genres of music.