Programming the beat: formalisation and modelling of techno music within an imitative musical system
The use of algorithms to compose music spans centuries and has progressed from manual calculations to computer-aided composition in the late 1950s, eventually forming the algorithmic composition (AC) field of research. Although Electronic Dance Music (EDM) has developed into its own musicological field of study, it is currently under-represented as a musical approach in current AC research. Despite the existence of music-theoretical analyses of EDM styles within academic literature (cf. Butler 2003, 2006; Zeiner-Henriksen 2010), a theory-based AC application of EDM does not exist in contrast to other musical styles (such as baroque music), while most EDM-based AC research mostly focus on data-driven approaches (cf. Anderson et. al. 2013, Marín 2018). This thesis aims to develop the foundational material for future techno-driven AC applications. The first half focuses on creating a formalised rule set from existing music-theoretical EDM literature and original musical analysis of a corpus of key techno tracks, and the second half describes the modelling and implementing this rule set in a system which can imitate the techno corpus. The two questions that this thesis aims to answer are (1) what are the musicological, low-level rules that govern the style of techno, and (2) How effective are these rules in imitating techno within a musical computer system.
This thesis is structured using Buchanan’s adaptation of the waterfall model. Identification consists of the literature review, followed by conceptualisation as the methodological outline. Formalisation of the rule set consists of the musicological analysis of techno music with examples available via the github repository and a content analysis to construct a corpus and sub-corpus of key techno tracks for different levels of musical analysis. The implementation of the system uses a hybrid software engineering methodological approach based on prototype evolution driven by a feature-based approach. An auto-ethnomethodological research log is used during practical sessions consisting of in-session note taking and postsession reflection. The testing stage consists of the evaluation of the rule set and system is by imitating the 10 tracks from the sub-corpus and evaluating the outputs both statistically and reflectively using the gathered reflexive material. A discussion of the findings and research itself is presented and highlights future developments of the system for both compositional and teaching purposes and potential future research including forms of automatic generation.