Siri Security Might Receive Boost with ‘Lexical Trigger’ Technology
Apple recently filed for a patent that reveals how Siri, the company’s personal voice assistant, could soon gain advanced functionality that will allow users to train her into responding only to their unique voices.
Presently, if enabled, Siri can be invoked directly from the iOS lock screen by simply holding down the home button on your compatible device. While certainly intended for convenience, this ease-of-access to Siri could potentially be a security threat if your device happens to fall into the wrong hands. For example, merely asking Siri “What is my name?” from the iOS lock screen will invoke the revelation of your personal contact information, including your name, address, and even photos. Likewise, Siri will respond to any command from anyone who’s able to make their way past the iOS lock screen, including your curious kids.
In the not too distant future, however, Apple may implement a new security-centric feature based on the concept of voice biometrics by which iPhone and iPad users will essentially be able to train Siri into learning and responding only to the unique sound of their voice. In other words, if Siri doesn’t recognize the sound of a user’s voice, he or she won’t be able to access their Siri-enabled iOS or macOS device.
The concept sounds simple enough on paper, but the underlying methodology, as noted by PatentlyApple, is actually a bit more involved, and could even be applied to work across the entire breadth of a user’s Siri-enabled devices.
As it stands, Siri may be invoked with a voice command, and will be responsive to the speech, itself, rather than to the speaker. While that’s certainly the point of a virtual assistant in the first place, the concept also entails that a user other than the owner of the device is able to utilize the voice assistant, which may not be desirable in all circumstances — for example, if your iPhone is ever lost or stolen.
Apple’s invention provides a multi-phasic method by which a user can train Siri to authenticate their unique voice before she’s able to respond to a given command. Referred to in the patent filing as the “Lexical Trigger,” users will essentially be able to customize the command to which Siri will respond. By default, for example, all a user has to do is say “Hey Siri” in order to invoke a response for her. However, in the future Apple wants to make it so that users can customize the command — for instance, “What’s good Siri?” — at which point she will only respond if the command and sound of the user’s voice matches up with what’s stored in Siri’s database.
The technology will also make it so that when a user wants to access Siri from their iPhone or iPad, they will need to be recognized on “multiple security levels” i.e. via Touch ID, advanced facial recognition, and the precise pitch and tone of their voice, otherwise she simply won’t be able to respond to commands or requests.