Comment by nfoz

Cool! Does text-to-speech require AI, or is there any active work in non-AI methods? Which bits are the AI bits? Do "deep" methods substantially improve over whatever classical methods we might have had?