Using inline tags in text to speech to alter pronunciation and other things

We're converting a weekly paper newspaper into speech, then to iTunes as mp3s to cut a CD or stick via Garage Band for a number of sight-impaired subscribers.

We've been using TextEdit to edit the text that we receive as pdfs (or copy/paste from the newspaper website) to make it suitable for listeners and of course shorter. Problem is, that some words are mispronounced, and we sometimes need longer pauses between paragraphs.

We get around the mispronunciation by mis-spelling words. That has been working for us, although Voice Over has exactly what we want from the pronuniation point of view, but does not work with the application we're using.

We've also worked out how to use some inline tags, such as [[slnc 5000]] but examples such as [[pron sym]]="h eh 1 l ow & w er 1 l d "[[/]] simply don't work. What am I missing? Do I need to go to a specific set of xml or other tags or a standard set of tags that always work with Apple's tts engine?

Look forward to a positive result.