I've just finished writing up my Physics PhD thesis and as part of the process, I had to proofread my work time and time again. I'm pretty bad at proofreading and kept missing stuff like "brunch charge" or "high energy bream" buried in the piles of text. I'd been using Grammarly and the like but still, things slipped through the cracks.
To help, I wrote a little Python function that uses Google's cloud text-to-speech engine to turn the .tex files into audio clips. You can find more detail on my blog or you can go straight to GitHub and check it out for yourself.
When said out loud, the typos and errors stuck out like sore thumbs. Any did oddly worded or overly long sentences were also much easier to spot. I also found the audio produced by Google's text-to-speech engine much clearer and easier to listen to than the options available locally on OSX. I can't recommend it highly enough.
If you're going to use this, remember that the cloud voices are free for 1 million characters a month, and $16 for every million characters beyond that. That's pretty pricey, so use it sparingly! You can use cheaper voices if you want, but they're not as listenable.
EDIT: to add more links
Source: reddit post