If you have closed captions, that should not be an issue, because letters map to braille characters.
No matter what you do, there'll always be some group who can't access the content.
E.g. people who are deaf, blind, and had both hands amputated.
I tend to agree with the other commenters that this seems like a good project for crowdsourcing because it's something many people do care a lot about, although I can see how that might not satisfy ADA compliance.
For all the talk about these types of issues, it seems to be very rare for tech conference videos to be captioned, see:
https://www.findlectures.com/?p=1&type1=Conference&talk_type...
I think this is potentially a good area to explore collectively - any technology solutions that make accessibility cheaper would feed back to places like UC Berkeley.
Do the captions need to meet a very high standard of quality? If not, automated captioning would have been a good student project.