UEDIN have developed a speech synthesis method which allows flexible manipulations of synthetic speech. It has potential as a tool for psychoacoustic investigations: to create natural-sounding perceptually-complete synthetic speech which is impossible using formant synthesis.
Goal: To develop existing techniques of articulatory-controllable statistical parametric speech synthesis into tools for psychoacoustic research, usable by other researchers, then demonstrate their effectiveness.
Relevance: New methods of stimuli creation will facilitate new experiments which are currently not possible.
Main host institution: University of Edinburgh
Second host institution: Universidad del Pais Vasco
Slides from Simon King's presentation in Sheffield, April 2013 (registered users only)