Researchers have shown a way to detect vibrations from a phone’s earpiece and figure out what the person on the other side of the call was saying with an accuracy of 83 percent, reveals a novel PTI report.
Unsplash
The tech, developed by researchers at Pennsylvania State University, is fabricated using an off-the-shelf automotive radar sensor as well as a state-of-the-art processing approach to show this rather threatening security concern.
The radar functions in the millimetre-wave spectrum in the bands of 60 to 64GHz and 77 to 81GHz, which was also the inspiration in naming their approach ‘mmSpy’. To the unaware, this is a subset of the radio spectrum that 5G uses.
In the tech demo, researchers simulated people speaking through the earpiece of a phone. The vibration of the earpiece via the speech moves across the body of the phone. This was then sensed by the radar and reconstructed to decipher what was being said on the other side.
This approach was effective even when the audio was inaudible to both humans. The sensory radar data is pre-processed via MATLAB and Python modules that help eliminate hardware-related, artefact noises from the data.
This is then fed to ML modules trained to classify speech and recreate the audio. From a foot away, the processed speech is 83 percent accurate. This drops down to 43 percent when the distance is at six feet.
Unsplash
Suryoday Basak, a doctoral candidate at Penn State, explains, “The methodology that we developed can also be used for sensing vibrations in industrial machinery, smart home systems and building-monitoring systems.”
“Imagine a radar that could track a user and call for help if some health parameter changes in a dangerous way. With the right set of target actions, radars in smart homes and industry can enable a faster turnaround when problems and issues are detected.”
For more in the world of technology and science, keep reading universo virtual.com.