Synchronous blind audio watermarking via shape configuration of sorted LWT coefficient magnitudes

This study explores the distributive characteristics of wavelet coefficients to enable blind watermarking in audio files. Binary embedding is performed in the low-frequency approximation subband on a frame-by-frame basis after the application of 3-level lifting wavelet transform (LWT). Within each frame, the sorted sequence of the approximation coefficient magnitudes is reshaped according to the intended bit rate of the watermark. A formula regulated by two controlling parameters is presented to carry out sorted sequence reconfiguration (SSR) without perturbing the magnitude order. To cope with time shifting and/or time scaling attacks, the proposed LWT-SSR scheme is further incorporated with an improved synchronization technique for the tracking of frame locations. Embedding strength, as reflected by the signal-to-noise ratio, is controlled at a level near 20 dB while operating at 86.13 bits per second. The objective differences obtained using the PEAQ metric indicate that the resulting watermarked audio signals are perceptually close to the original. Comparison results confirm the robustness of the proposed LWT-SSR against common signal processing operations as well as time-scaling and time-shifting attacks.