AT&T; Labs’ text-to-speech demo is better than any I’ve ever heard. I mean, the Bay Area’s NPR affiliate has announcers who don’t sound this natural. I’m willing to believe, on the strength of this, that this time next year, I may be making “audiobooks” out of a bunch of Web clippings and ASCII and just running them through a TTS widget like this. Link Discuss (Thanks, Elias!)