Impressive next step for text-to-speect. Wish there was some simple real demos. I also work on the same thing using DL- and hope to open source the "emotional part" of it.
We soon can create emotionally expressive youtube videos with synthetic actors..