Interesting! Deep learning for text-to-speech synthesis