Interesting idea and in research we do it when we have authorization from users to do so. There are tools out there to do this (glance, ethnio, etc), but nothing as classy as what goggle is coming up with..
http://googleresearch.blogspot.com/2006/06/interactive-tv-conference-and-best.html#links
"We showed how to sample the ambient sound emitted from a TV and automatically determine what is being watched from a small signature of the sound -- all with complete privacy and minuscule effort."
A little ambient evesdropping eh? Now thats tasty!