No real meat here. Probabaly a good idea that was not communicated effectively. Dunno what you are really talking about specifically.

Hennath, it seems to me to be a pretty simple idea, even if it's doable or not.

teknet wants to follow someone's lips with a camera and appropriate software and see it the computer can read their lips, like a deaf person would.

I know much study has gone into the 3d production of sounds for use in CGI people, so it's quite clear that computers can produce correct mouth and lip placement to make it look like a CGI person is talking. But can they recognize them?

I think the biggest drawback is that the computer would probably not be able to get enough data from a 2d image of a person talking--likely to recognize the speech, a computer would need a 3d scan. But again, doable. I don't know if it's valuable--right now if you want to read lips, just get a deaf person.

It's doable. Sometimes we want to do something ourselves, no need a deaf person to read lips for us. :). Thanks for the contributions. Welcome criticism of all kinds.