Researchers Create Eavesdropping Attack For Android Devices That Detect Caller’s Identity

A new academic effort has come forward thanks to a team of researchers that have created a unique eavesdropping attack for Android devices.

The attack is designed to determine a caller’s gender and identity and even go about outlining private speech so it’s actually eavesdropping through the likes of motion sensors that detect movement and reverberations through ear speakers located across mobile devices.

The new creation is called EarSpy and it’s a joint effort from researchers located at five different US-based universities.

This kind of attack is greatly explored in loudspeakers located across smartphones, some of these ear speakers were seen to be too weak to produce enough vibrations to attain a risk for eavesdropping that turns side channel attacks into something more practical.

In the same way, it’s important to understand that modern-day devices make use of powerful kinds of stereo speakers as compared to different models that arose a few years back. They end up producing sound quality that’s so much better and with more enhanced vibrations.

And then you’ve got modern-day devices that make use of sensors that are more sensitive in design and a few gyroscopes which end up recording small resonances through such speakers.

Now the proof of how much progress is actually done is revealed through the project’s experiment. They’ve provided a detailed overview of how it was created and what results were produced in this regard.

It’s all a combination of apps from third parties and devices like the OnePlus 9 and also the likes of the OnePlus 7T, along with different sets of audio that had been pre-recorded and played solely via ear speakers of at least two different types of devices.

Other than that, we have an algorithm that makes use of machine learning and it had been trained in a ready manner for various datasets to highlight speech content, gender, and the identity of callers.

All the data from this test was variable and it depended on the device and the dataset seen but it ended up producing some great results in terms of eavesdropping through an ear speaker.

Thanks to an app called Spearphone, the researchers created during a trial to counter the attack from 2020. Its accuracy is great and we’re talking gender as well as ID accuracy which totaled out to be 99%. Meanwhile, the accuracy for speech recognition was a good 80%.

Read next: New Study Says Most Buyers Blindly Trust Social Media Influencers For Purchase Decisions
Previous Post Next Post