If you have Android 2.2, you may have noticed an update to Google Voice Search. The big change is an opt-in service that allows Google to help recognize your patterns of speech. Like Dragon Naturally Speaking and other speech recognition software that have come before, you can train the speech engine – in this case Google – to recognize the nuances of your dialect or manner of speaking.
If you’re not familiar with Android’s speech recognition, you speak to your phone and the waveform created gets transmitted to Google, analyzed and sent back to your device. This means for those who opt in to this service, your voice and your speech patterns will be sent to Google. How’s that for privacy issues?
How do we protect the data?
The ‘electronic keys’ are designed to be accessed by machines. Very few people within Google, who passed careful vetting, will have access to them. The personalized acoustic and language models are binary files designed for use by machines.
So the data is encrypted both ways and at rest. It can only be accessed from a physical device, not by individuals searching around the data center. And they’ve put the human factors back into it, with special vetting procedures for those “very few” who will have access to these special machines.
If you change your mind about participating, Google says you can go into your account, turn it off and they will immediately destroy the keys, thereby destroying any way to access the data. They will also destroy the link between your device and the files and the data files, themselves.
Thoughts? Comments? Will you sign up for this?