Many websites use tests intended to distinguish humans from automated processes as part of account registration and other functions. Widely known as Completely Automated Public Turing tests to tell Computers and Humans Apart (CAPTCHA), such tests are generally intended to supply a visual or audio task that is relatively easy for humans but hard for computers. In this paper, we propose a generic approach for breaking audio CAPTCHAs based on advanced audio processing and machine learning algorithm. We show that our system is able to break all the popular audio CAPTCHA schemes, including Microsoft and Yahoo, that use non continuous speech.
The failure of noise-based non-continuous audio captchas
|Security and Privacy
|Elie Bursztein , Romain Bauxis , Hristo Paskov ,