Audio authenticity is one of the major tasks for audio forensic experts because it is often a requirement for the admissibility of digital audio evidence. This investigation proposes a passive authentication method based on an adaptive audio fingerprinting scheme to detect forgeries produced by the replication of an audio interval within the same evidence. Several audio fingerprinting systems are analyzed, and an adaptive scheme based on the Fourier spectrum distribution is proposed. The adaptive system is theoretically and empirically adjusted to detect short replicas. Simulations are performed to analyze the robustness against time, frequency domain, and compression distortions. The method has the power to discriminate repeated text speech and distinguish it from audio replicas as short at 0.1 s even in the presence of amplitude and frequency distortions.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.