This is great. I like the idea of text-to-speech, but rather than do it yourself, why not partner with Stitcher? They already convert content (video and text) into audio for listening through their mobile app, and now they're bundled (along with Pandora) into 2010 Ford Sync II equipped vehicles, so it seems like a good fit.

Disclosure: I did some usability work for them last year, and they're the nicest bunch of people :) (I'll gladly introduce)