Sign up for our free weekly e-newsletter

And how would this technology work? It would “sample” the ambient background sounds from your television, through the microphone in your PC or laptop while you’re surfing. So, you could be watching your favourite team while idly looking through potential travel destinations, when up pops a promotion for tickets to the next home game. The technology is described in a paper http://www.mangolassi.org/covell/pubs/euroITV-2006.pdf freely distributed by Google.

Now, if we all listen carefully ourselves, I’m sure we can hear the ambient background noises of a thousand privacy advocates gearing up to take Google to task for even considering this! But before we leap to the conclusion that Google has figured out how to turn your computer into an eavesdropping device, we should step back a second and remember that “sampling” is not necessarily the same as “listening in.” Technically, sampling is an electronic technique used to convert an analog signal to a set of digital bits and bytes at various points in time. If you sample very rapidly, you can convert an analog sound signal to a digital one and put it on a CD instead of a tape or a vinyl record, with no real loss in quality. But if you sample less rapidly, or if you sample differently in some way, then what you get are a string of data points that can be used to match this signal to other known signals, and this is how Google’s potential application could identify the television channel you have playing in the background.

You don’t need a lot of data points to make this determination – not nearly as many as you would need to reconstruct the actual sound. Remember the brouhaha about “cameras” that scan faces in a crowd for security purposes, matching them with the known faces of terrorists? The devices don’t actually take real “pictures” of people, but rather they rapidly “sample” the data on the many faces in a crowd, computing various lengths and facial ratios – like the ratio between the distance from eye to eye and the distance from chin to nose, for instance. A few data points on each face are all that is needed to do a precise match, at which point, presumably, the terrorist’s real photograph would be called up from another database.

The Google-sponsored research paper proposing the acoustic sampling software maintains that the data stream from a computer’s background sound would be “irreversibly compressed” before being sent on to Google’s servers – implying that it cannot then be used (i.e. listened to) in any other format or for any other purpose. According to the paper, “The viewer’s acoustic privacy is maintained by the irreversibility of the mapping from audio to summary statistics….[O]ur approach will not ‘overhear’ conversations. Furthermore, no one receiving (or intercepting) these statistics is able to eavesdrop on such conversations, since the original audio does not leave the viewer’s computer and the summary statistics are insufficient for reconstruction. Further, the system can easily be designed to use an explicit ‘mute/un-mute’ button, to give the viewer full control of when acoustic statistics are collected for transmission.”

Still, the very fact that Google, or any other technology-based online service for that matter, can actually access the sounds coming in through your computer microphone is something I think most of us find, well, creepy. And while the company would of course tell you they aren’t interested in eavesdropping, how do you prevent it from happening when this kind of software falls into the hands of less scrupulous users? In the future, will we have to disable our microphones before going online?