Princeton University researchers conducted an online tracker study revealing a new technique, called audio fingerprinting, that removes the privacy of Internet users through their computers’ audio stacks. This technique uses a software, Audio Context API, that fingerprints a device by collecting information about its audio signature to track the web user. The researchers compare audio fingerprinting to canvas fingerprinting, which also collects browser data to track user activity; both methods use software to harvest a digital fingerprint of devices and user behavior.

While not yet ubiquitous tracking techniques, audio fingerprinting and other little-known methods can evade successful tracker tools like Ghostery and Firefox’s cookie blocker. This approach poses a problem because many fingerprinting scripts go unnoticed by these privacy tools, opening up user information to trackers. The Princeton researchers foresee machine learning as a way to combat fingerprint scripts in the future. Through detection and classification of trackers, machine learning will eliminate manual curation of tracker blocker lists, efficiently improving the privacy of browsers.