Sound can be used to detect many kinds of events, like medical symptoms, usage of kitchen appliances, and doors being opened. However, nobody wants an always-on microphone that is able to capture speech when the system gets compromised. A consumer microphone will only capture sound within the audible spectrum, so filtering out the range of speech leaves too little information to work with.
This study shows that infra- and ultrasound also works well for detecting significant events. PrivacyMic uses hardware filtering to remove the range of speech, so it cannot be used to gain sensitive information even if an attacker had full control of the software.
This study shows that infra- and ultrasound also works well for detecting significant events. PrivacyMic uses hardware filtering to remove the range of speech, so it cannot be used to gain sensitive information even if an attacker had full control of the software.