i personally really like the idea of speech synthesis.. when i hear a document i remember it much better.
i found that speech synthesis in Linux is a little problematic. festival is very strong on commands, but i can't understand what it says, it sounds like a child that doesn't know yes where to breath in a sentence.
i do know of people that have to use festival, because they have no other choice, you can visit speak up at: http://www.linux-speakup.org/
this is a project made, as they say in the site, by a bunch of blind people that like to mess around with Linux. but i personally never made contact with them.
another direction i thought towards is KDE, finally on version 3.5 the KTTD (text to speech daemon) started working out of the box, and kaysayit might be a widely used tool.
that is my experience with festival