Text to speech is an ability of some computing devices to run a special program that will "speak aloud" a given text to you. That is, they process the text, and generate the sounds that resembles a person speaking that text aloud using speech synthesis algorithm.
A lot of different software supports text to speech; among them, Microsoft Reader as well as Kindle - at least the latter software running on Kindle 2d-gen or DX - here are instructions.
The difference from an audio book is that the latter is just an audio file; a recording of an actual human reading the book aloud, not the program producing the sounds.
Think of it as analogous to the difference between a video file from a movie depicting an actor doing things; and a computer game demo where a rendered video game character does those things on-screen.
"text-to-speech: Enabled" means that the book supports the ability of supporting book reader to render its text into speech, both from technical and licensing perspective. In case of Kindle Amazon books, since the technical capability is always there (same format), it's merely a question of licensing - see discussions of this (somewhat controversial and unpopular) feature here and here.