Researchers from the U.S. and China have been able to digitally re-create fingerprints from analyzing the sounds of said fingers using a touchscreen. Called PrintListener, the technique involves listening to the sounds created by the friction of moving a fingertip against a touchscreen surface.
When tested in real-world scenarios, PrintListener was able to successfully re-create both partial and complete fingerprints from sounds in the background of Discord and FaceTime calls. After isolating the sounds from the background of the calls, the system then uses minimum redundancy maximum relevance (mRMR) alongside an adaptive weighting strategy to account for how behavioral and physiological factors may have influenced the sounds.
It uses all these data points to detect the features of the left loop, right loop, and the whorl of a fingerprint and then produce a synthetic re-creation. It was successfully able to fool the Automated Fingerprint Identification System in 1 in 4 attacks with partial fingerprints and 1 in 10 attacks with complete fingerprints.
