Voice-activated devices such as smart speakers, smart displays, smartphones, etc. have been shown to be susceptible to attacks that utilize directed laser beams to activate the device. Such attacks can cause the device to respond to queries even in the absence of actual spoken commands from a user. This disclosure describes multiple approaches such as the use of a physical barrier and use of signal processing techniques to analyze signals detected at multiple microphones of a device to mitigate such attacks. The device can be designed with a light blocking geometry to minimize visible light that reaches the microphone membrane. The device can also include a light sensor in the vicinity of the microphone to detect the presence of light inside the microphone port. Output of the light sensor can be used in the device security logic.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.