Authors:
Salma Masmoudi
1
;
Maha Charfeddine
1
and
Chokri Ben Amar
2
Affiliations:
1
REGIM: REsearch Groups on Intelligent Machines, National Engineering School of Sfax (ENIS), University of Sfax, Sfax 3038, Tunisia
;
2
Department of Computer Engineering, College of Computers and Information Technology, Taif University, Taif 21944, Saudi Arabia
Keyword(s):
Audio Watermarking, Tamper Detection, Authenticity, Integrity Control, Recovery.
Abstract:
Audio recordings contain very sensitive content, such as historical archival material in public archives that protects and conserves our cultural heritage, digital evidence in the context of law enforcement, the online formats of sensitive digital Holy Quran, etc. Such audio content is vulnerable to doctoring and falsification of its origin with malicious intent. One tool to solve several multimedia security difficulties facing this sensitive content is to tag it with a message before the distribution process. This technique is called watermarking. Hence, this paper aims to present a scheme of tamper detection and integrity control based on multipurpose and secure audio watermarking. To treat the integrity control application, we suggested embedding in the digital audio signal the tonal components resulting from the Human Psychoacoustic Model masking study, which are extracted as features from the relevant low-frequency band of the original audio signal. In addition, a Multilayer per
ceptron-based denoising autoencoder was executed after learning robust representation from corrupted audio features to correct the watermarked frequencies, thereby restoring the original ones. Consequently, blind tamper detection and blind invertibility were guaranteed. The detailed results indicated that the suggested scheme achieved higher performance at the integrity control and tamper detection level, as well as at the watermarking and reversibility properties.
(More)