It would probably be most effective (as opposed to the most easy) to parse the HTML, and extract the main text; there's also another option, there exist scripts for producing an epub from fanfiction.net (amongst other websites), which would probably lend itself well to being run through a TTS engine...