Synthesising Consistent Instrument and Voice Sounds using Physically Informed Spectral Models and Evolutionary Learning
In collaboration with the University of Limerick (Dr. Niall Griffith & Dr. Jacqueline Walker) we are applying physical modelling methods in combination with neural and evolutionary learning to improve the quality of synthesised sounds. Variation in the sound components over a musical instrument's range reflects its class, e.g. woodwind, brass, string. Current approaches do not address adequately the issue of consistency over the range of an instrument. The aim is to develop models that facilitate the mimicry of existing musical instruments, the design of new instruments that are consistent with a class, and new hybrid classes of instrument. This question will be addressed by developing models that reflect the characteristics of physical sound sources but which provide control of synthesis in terms of their perceived characteristics rather than low-level parameters.

Elevated Pitch: A Grammatical Genetic Programming Approach to Group Based Composition of Contemporary Music
John Reddin undertook this research as an intern at the Online Dublin Computer Science Summer School in the Summer of 2008. The work explored the semi-automation of music composition using Grammatical Evolution, and the effect of group collaboration on the compositions. The Elevated Pitch software was developed with GEVA.

An Interactive Evolutionary Algorithm for Sound Synthesis
In collaboration with Dr. Niall Griffith and James McDermott (University of Limerick) we developed a novel Interactive Evolutionary Algorithm for Sound Synthesis. This project was funded under an IRCSET Embark Postgraduate award.