Rumours of a Google TV product have been flying around the web for a while, with the latest pointing towards some kind of huge searchable programme guide, which would then let you stream the content you find to your TV. [via Google Blogoscoped]

Well, some new clues have emerged about a new direction the Google TV offering could take. On the Google Research blog the latest post talks about a paper they presented at the Euro ITV interactive television conference. In it they talk about a system that uses the microphone on your laptop or desktop PC.

The microphone picks up the sound coming from the TV, your laptop processes it and sends a statistical audio fingerprint to the "Google Audio" server. Google identifies the audio, and sends back relevant information about what you are watching. This could be websites, news items etc. It can also link you with people watching the same thing, creating an instant chat community. It would also be able to provide live viewing figures, and of course display ads relevant to what you are watching! All this is independent of any television technology or format, now or in the future - very clever.

Here are some key snippets from the paper.

"..while watching a news segment on Tom Cruise, a fashion layer might provide information on what designer clothes and accessories the presented celebrities are aearing." - A huge meta data surrounding broadcast material will be needed for this?

"Thus, content providers or advertisers might bid for specific television segments. For example, local theaters or DVD rental stores might bid on audio from a movie trailer.." - Adwords for TV then!

"the closed captions provide keywords useful for searching for related material. The search results can be combined with a viewer's personal profile and preferences (ZIP code and 'fashion') in order to display a web-page with content automatically obtained from web-pages or advertisement repositories using the extracted keywords." - An automated Google Local based on what you're seeing?

"Importantly, as the viewer's viewing context changes (by changing channels)...re-sampling the ambient audio...the viewer need never indicate what program is being watched;" - Clever, as you change channels, it updates conversations, information, ads, to what you are watching without any input from the user.

"the popularity ratings are easily derived by simply maintaining counters on each of the shows being monitored." - Realtime audience figures, which leads to..

"advertisers and content providers to dynamically adjust what material is being shown to respond to drops in viewership." - clever, now their data is of value to advertisers.

The last idea they have is bookmarking what you're watching for future reference with a single click, they'll figure out what it was from the audio fingerprint at that time. Then from that they can direct you to view that programme online (using Google Video) at a later stage, if its available online.

The rest of the paper goes into technical details about how they do the audio matching, but it seems they've run experiments on this. So what does this all mean? Well in my opinion I think this could be where Google TV is heading, this is the biggest set of clues they've let out of the bag yet. I'd also think it means it is not going to be released anytime soon.

But what do you think, the paper is openly available, take a read and see what conclusions you can draw from it.