Based on a few titles I checked, the text-to-speech sounds very natural. There was a hiccup or two here or there, but at about the hit rate of "I would have done that differently if I were reading it out loud" I get from human-read stuff.