AES E-Library

Applications of Modified Discrete Cosine Transform to Detect Audio Copy-Paste Fragments

Different international scientific organizations developed best practice guidelines and manuals for forensic authentication of digital audio recordings [1, 2] that contain frameworks to analyze and authenticate digital audio recordings. Copying-pasting, inserting, and cloning audio samples are editing techniques that can change the meaning of a recorded conversation and alter the authenticity of the initial recording. Previous work [3, 4, 5] reports on applications of Mel-Frequency Cepstral Coefficients (MFCC), Local Binary Pattern (LBP), Discrete Fourier Transform (DFT), Short-Time Fourier Transform (STFT), Content-Based Copy Detection (CBCD), and partial matching without query. This paper is part of a larger research project regarding the Modified Discrete Cosine Transform (MDCT) coefficients and Map [6] analyses for forensic purposes, and introduces a new method, MDCT-based, without query to detect cloned audio fragments. The paper presents: (1) applications of MDCT to analyze the artifacts introduced by copy-pasting audio fragments; (2) the robustness of the proposed method against editing techniques like inverting, amplifying, and attenuating the paste fragments; (3) the robustness of the proposed method against added noise; (4) the robustness of the proposed method against lossy audio re-compression; (5) objective methods to report the comparison results; (6) the actual limits of the method and future research. Future research will present extended results and applications of the proposed method. With these findings, the use of the proposed method in forensic audio analyses together with other scientific validated methods is recommended.

 

Author (s):
Affiliation: (See document for exact affiliation information.)
Publication Date:
Session subject:
Permalink: https://aes2.org/publications/elibrary-page/?id=22627


(2014KB)


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 Join the AES. If you need to check your member status, login to the Member Portal.

Type:
E-Libary location:
16938
Choose your country of residence from this list:










Skip to content