Inaudible ‘shadow’ sound technique promises new security, communications applications

Inaudible ‘shadow’ sound technique promises new security, communications applications

Technology News |
In a paper titled "BackDoor: Making Microphones Hear Inaudible Sounds" presented at the MobiSys 2017 conference held in Niagara Falls last week, researchers from the Coordinated Science Laboratory at the University of Illinois have devised a way to create microphone-only audible sounds out of encoded high-frequency sounds.
By Rich Pell


While sounds beyond the 20kHz audible range can’t be detected either by the human ear or by standard microphones, high-frequency sounds beyond a microphone’s typical recording range can be designed to exploit non-linearities in microphone hardware (its non-linear diaphragm and power-amplifier) and create a “shadow” in the recordable (microphone-audible) frequency range.

That “shadow” signal could even be regulated to carry data bits, explain the researchers in their paper, opening an acoustic (yet inaudible) communication channel via today’s microphones. The researchers envisage many applications, including unique IoT communication channels (up to 4kbps at 1m) relying on ultrasound emission rather than on Bluetooth radio, and audio jamming solutions that would help secure private conversations in an untrusted environment.

The researchers dubbed their technique “BackDoor” since it can be readily applied with any microphone available on the market, from smartphones to hearing aids, without any modification at the receiver side.

To yield the “shadow” sound, the frequency and phase of sound signals must be designed so that when they pass through the non-linear amplifier of the microphone, the high-frequency sounds then create a low-frequency signal within the filtering range of the microphone, which gets recorded as normal sounds. In its most basic implementation, consider BackDoor plays two tones at two well-separated ultrasound frequencies (f1 at 40kHz and f2 at 50kHz).

Once those reach the microphone, they not only get amplified buot also multiplied due to fundamental non-linearities in the system. It is the multiplication of the two frequencies that yields frequency components including (f1+f2) but also (f1-f2) (in this case 10kHz) which is within the microphone’s recording range.

Stepping up their game, the researchers demonstrated that data could be loaded on transmitted ultrasound carrier signals so the resulting “shadow” signal could then be demodulated after recording. In a jamming scenario, the ultrasound frequencies could be played so as to generate a white noise that would predominantly be picked up by any hidden microphones over the conversational voices.

For more, see “BackDoor: Making Microphones Hear Inaudible Sounds” (PDF).

Related articles:
Audio control energizing microphone, speaker markets
Home audio market: ‘It’s all about the voice’
Far-field voice capture demonstrated using radar, audio beamforming
MEMS microphone technology adapting to changing use cases

Linked Articles