There's a massive business for audio correction on the web. I find it almost hysterical to have a 1080p conference video with a blurry slide and a dark speaker and cannot hear clearly. Radio is a better medium.
I’m interested in presentation detection and enhancement in general. For example, if a person is standing in front of a mostly static chalkboard for an hour, there is a ton of information there about the writing on the board. The frames could be fed into a handwriting detector and perhaps vectorized, and the video format should understand this and transmit a full resolution chalkboard on frame zero of the transmission, and not send any board info until there is an actual modification.