Defending Against Microphone-Based Attacks with Personalized Noise

Abstract Voice-activated commands have become a key feature of popular devices such as smartphones, home assistants, and wearables. For convenience, many people configure their to be ‘always on’ listening for voice from the user using trigger phrase “Hey Siri,” “Okay Google,” or “Alexa.” However, false positives these triggers often result in privacy violations with conversations being inadvertently uploaded cloud. In addition, malware that can record one’s remains signifi-cant threat privacy. Unlike cameras, which physically obscure assured privacy, do not way knowing whether microphone is indeed off are left no tangible defenses against based attacks. We envision general-purpose physical defense uses speaker inject specialized obfuscating ‘babble noise’ into microphones protect automated human present comprehensive study how specially crafted, personalized ‘babble’ noise (‘MyBabble’) effective at moderate signal-to-noise ratios provide viable eavesdropping