Researchers from Google labs have tested a system for what they call mass personalisation, based on listening to ambient audio from a television to identify programmes and present relevant information on a web browser.

Their study, on “Social and interactive television applications based on real-time ambient audio identification,” received the best paper award at the recent EuroITV 2006 conference in Athens.

They describe a system that uses a personal computer with a built-in microphone to listen to ambient audio and build identifying signatures that are sent to a central database. It requires no additional hardware and works with any television system, even during conversations between viewers.

The researchers claim that user privacy is maintained because it is not possible to covert the signatures back into audio and users remain in control over when data is collected. The system requires only a few hundred bytes per second to identify the audio of programmes or advertisements.

The approach is similar to some ‘people meter’ systems used to measure radio or television usage for ratings, or systems that allow music tracks to be identified by their audio.

“We showed how to sample the ambient sound emitted from a television and automatically determine what is being watched from a small signature of the sound — all with complete privacy and minuscule effort,” said Google researchers Michele Covell and Shumeet Baluja. “All of this would be done without users ever having to type or to even know the name of the programme or channel being viewed.”

The Google researchers describe four applications to make television more personal, social and interactive: providing additional relevant information or advertising; forming ad hoc communities; generating real-time popularity ratings; and enabling television-based bookmarks.

www.google.com