There is new software that can listen in on conversations, identify the gender and identity of the caller, and even understand what is being said to some extent.

The good news is that the virus was created as part of a white hat experiment and presents no threat to smartphones (at the time).

Experts from Texas A&M University, the New Jersey Institute of Technology, Temple University, the University of Dayton, and Rutgers University collaborated to develop EarSpy.

Using the equipment improperly

Side-channel assault EarSpy takes use of the improved microphones, accelerometers, and gyroscopes found in modern smartphones.

As the endpoint’s ear speakers echo during a discussion, the malware attempts to access the data gathered by motion sensors. Due to less robust speakers and sensors, this wasn’t a practical attack vector in the past.

The researchers utilised a smartphone from 2016 and a smartphone from 2019 to illustrate their argument. It was plain to see the increase of information collected.

The data was tested on a OnePlus 7T and a OnePlus 9 to see whether it could be used to determine the caller’s gender and recognise their voice.

The former had a caller gender identification rate of between 77% and 98%, while the latter had a rate of between 63% and 91%. The accuracy of speech recognition varied from 51.8% to 56.4%.

Whitepaper authors claimed that despite “10 distinct classes,” accuracy was five times better than a random estimate, suggesting that ear speaker vibration had a significant effect on accelerometer readings.

While researchers had a high rate of success (average of 88.7%) in determining the gender of a caller using the OnePlus 9 smartphone, they were only able to identify the caller (on average) 73.6% of the time. There was a decrease in accuracy of speech recognition of 33.3% to 41.66%.