This works very well, although there are some areas for improvement:

+ Its parser reads all the text of site elements. This means that texts like "Login, Navigation" are also read.

+ It might be better to use just-in-time requests rather than generating all possible site text before sending requests, which would force the user to wait a while before listening to the site.