Creativity, exploration and control in musical parameter spaces

This thesis investigates the use of multidimensional control of synthesis parameters in electronic music, and the impact of controller mapping techniques on creativity. The theoretical contribution of this work, the EARS model, provides a rigorous application of creative cognition research to this topic. EARS provides a cognitive model of creative interaction with technology, retrodicting numerous prior findings in musical interaction research. The model proposes four interaction modes, and characterises them in terms of parameter-space traversal mechanisms. Recommendations for properties of controller-synthesiser mappings that support each of the modes are given. This thesis proposes a generalisation of Fitts’ law that enables throughput-based evaluation of multi-dimensional control devices. Three experiments were run that studied musicians performing sound design tasks with various interfaces. Mappings suited to three of the four EARS modes were quantitatively evaluated. Experiment one investigated the notion of a ‘divergent interface’. A mapping geometry that caters to early-stage exploratory creativity was developed, and evaluated via a publicly available tablet application. Dimension reduction of a 10D synthesiser parameter space to 2D surface was achieved using Hilbert space-filling curves. Interaction data indicated that this divergent mapping was used for early-stage creativity, and that the traditional sliders were used for late-stage fine tuning. Experiment two established a ‘minimal experimental paradigm’ for sound design interface evaluation. This experiment showed that multidimensional controllers were faster than 1D sliders for locating a target sound in two and three timbre dimensions.