Model Talker

Researching how to create the most expressive form of my own synthesised voice


Via a really-helpful stranger in the MNDA called Richard Cave (himself via a chain to two wonderful people in the NHS – from my MND Coordinator, Tracy Thomas, to my Speech and Language Therapist, Caroline Ayers) I reached Tim Bunnell, the Head Research Scientist of the Speech Research Lab at Nemours [the US team responsible for developing ‘Model Talker’, a Text-To-Speech (TTS) system that can be taught to speak in your own voice ]. He has agreed to use me as a guinea pig to research putting ‘expressiveness’ into TTS – rather than it sounding like Robot-Peter on Tranquilisers. To start the research, I have to record 3,200 rigorously-specified phrases, carefully matching the pace of a template example. It’s taking a little bit longer than expected because the lab computer in the USA that I link to in real-time keeps telling me my pronunciation is off – basically because I’m not speaking American-English. But I find the potential of this research really exciting!