Atomic Structure Parsing: A Novel Framework for Digital Audio file Forgery Detection

Abstract

Digital advancements, accessible free editing tools, and Artificial Intelligence have fundamentally transformed multimedia authentication challenges. Digital audio files being the primary medium of communication and crucial evidentiary material in the court trials, the authentication and the integrity check of the digital audio files is necessary to rule out any intentional forgery or tampering in them. The blind detection of the tampering in the absence of the recording media or reference audio file, is in itself an another major challenge which is being addressed through the method discussed in this paper. The paper presents an in-depth insight into the structure of the digital audio files, forgery detection and ultimately aims at determining the authenticity and thus integrity of the audio files based on the atomic structure parsing and analysis of the container of digital audio files and exploiting the metadata and the structural values of the container atoms thereof. The proposed method works efficiently to identify the tampered digital audio files altered by the means of cutting, copying, copying-pasting and deliberate noise addition modifications. By detecting post-manipulation re-indexing and re-muxing through container structure anomalies of addition, deletion or changes in the patterns in the atomic structure, the proposed approach empowers the forensic examiners to reliably differentiate authentic primary digital audio recordings from secondary or tampered derivatives for court proceedings.