It looks very promising. What text-to-speech engine are you using?