Text this: Voice Quality Modelling for Expressive Speech Synthesis