Hybrid Blind Audio Watermarking for Proprietary Protection, Tamper Proofing, and Self-Recovery

The paper presents a lifting wavelet transform (LWT)-based framework for multi-purpose blind audio watermarking. The proposed schemes can be used to carry out robust watermarking for intellectual property protection as well as fragile watermarking for tamper detection and signal recovery. Following 3-level LWT decomposition of the host audio, the coefficients in selected subbands are partitioned into frames for watermarking. To expand applicability, the robust watermark comprising proprietary information, synchronization code, and frame-related data was particularly embedded in the approximation subband using perceptual-based rational dither modulation (RDM) and adaptive quantization index modulation (AQIM) at a payload capacity of 1523.9 bits per second. The fragile watermark is a highly compressed version of the audio embedded within the 2^nd- and 3^rd -level detail subbands using 2^N - ary AQIM. Hashing comparison and source-channel coding make it possible to identify tampered frames and restore affected regions. Experiment results indicate that the embedded robust watermark can withstand commonly-encountered attacks and the fragile watermark is highly effective in tamper detection and self-recovery. More importantly, the incorporation of a frame synchronization mechanism makes the proposed system resistant to cropping and replacement attacks, all of which were unsolvable using previous watermarking schemes. The perceptual evaluation revealed that the watermark caused only minor degradation. The proposed watermarking scheme is suitable for a wide range of ownership protection and content authentication applications.