UES256—Upper Euphrates Signature: A Secure Fragile Watermark for Blind Authentication and Tamper Detection
Ahmed A. AlSabhany, Mahmood Abdulrazzaq Alsaadi · IEEE Access · 2025
Fragile watermarking provides integrity verification and authenticity assurance by embedding tamper-sensitive watermarks to reveal unauthorized modifications. In the audio fragile watermarking literature, many existing methods exhibit limitations in tampering sensitivity, watermark coverage, and security against unauthorized detection, removal, and forgery. This paper presents a blind, secure, and highly sensitive fragile audio watermarking method designed for signal authentication and tamper detection. The method integrates SHA-256 hashing, block-level key generation using a cryptographically secure pseudorandom number generator, and Hamming syndrome embedding. This design achieves full-signal watermark coverage, preserves perceptual transparency, and enforces strong watermark security. The watermark serves as a cryptographically bound signature that is verifiable but cannot be reconstructed or forged without the original signal. To rigorously evaluate its tampering sensitivity, the method was tested against randomly flipped single bits distributed across the entire signal. It achieved a detection accuracy of 99.7% while maintaining high perceptual quality at 103 dB SNR. Leveraging Hamming syndrome embedding, the method compresses the watermark into a 256-bit payload, enhancing undetectability and preserving signal fidelity. Additionally, it supports tamper localization at both block and bit-plane levels, achieving 99.7% and 88.6% accuracy, respectively. A comparative evaluation against six existing fragile audio watermarking methods demonstrates the superior performance of the proposed approach in tamper sensitivity, watermark coverage, perceptual transparency, and security. Finally, the method operates in the temporal domain and therefore is applicable to authentication, forensic validation, and copyright protection across audio, images, and other redundancy-rich signals.