This Project's

Research Team

Co-ordinators

Students

  • Matt Smith

Artificial intelligence Melody and Education

Theme: Music Computing
This project laid the groundwork for the development of intelligent learning environments for novice composers. One goal was to find ways to use artificial intelligence (AI) to facilitate melody composition. The research had two stages. The first was the formalisation and testing of an existing analytical theory of melody: Narmour's implication-realisation theory. This theory offers an explanation of how musical listeners break a melody into 'chunks' and hear some notes as more important than others. The formalisation involved the implementation of a parser to create hierarchical analyses. A critical evaluation of the theory and the parser was based on comparison of published analyses based on the theory with those created by the parser from the same melodies. The second stage of the research involved extending the parser with constraint-based generation techniques. One result was an AI tool, MOTIVE, which can generate melodies given an analysis (either from the parser, or constructed by a user) and a set of constraints to be applied at each hierarchical level. Publications: M. Smith & S. Holland, (1994) “MOTIVE: A constraint-based tool for melody analysis and generation”, In M. Smith, A. Smaill & G. Wiggins (eds), “Music Education: An Artificial Intelligence Approach”, Springer-Verlag M. Smith & S. Holland, (1993) “MOTIVE: A constraint-based tool for melody analysis and generation”, Proceedings of Workshop on Artificial Intelligence, Music and Education, part of AIED-93, Edinburgh, Scotland M. Smith & S. Holland, (1992) “An AI tool for the Analysis and Generation of Melodies”, Proceedings of International Computer Music Conference, San Jose, California, USA M. Smith & S. Holland, (1992) “An Intelligent Tutor for Melody Composition”, Proceedings of International Conference on Systems Research, Informatics and Cybernetics, Baden-Baden, Germany