May want to look into these papers

http://speechprosody2010.illinois.edu/papers/100096.pdf http://www.springerlink.com/index/P32774KN22650J42.pdf http://www.softsea.com/review/Pistonsoft-Text-to-Speech-Conv...

I've heard that adding breaths into the mix are not just for the speaker, but that the listener can take that moment to mentally index what the speaker has just said. I can't find the paper at the moment, but there's been some good research that simply adding breathing sounds and pauses at key points in the audio can make listening to long TTS passage more bearable.