Digital audio watermarking using minimum-amplitude scaling on optimized DWT low-frequency coefficients

This work presents an audio watermarking scheme using minimum-amplitude scaling on the optimized lowest-frequency coefficients in the wavelet domain. Signal-to-noise ratio (SNR), bit-error-rate (BER) and Perceptual Evaluation of Audio Quality (PEAQ) are commonly utilized performance indexes in measuring the fidelity, robustness and quality of watermarking algorithms. However, there is a tradeoff relationship between audio quality and robustness. To overcome the drawback, this paper aims in proposing an optimization-based scaling scheme using optimal modification of low-frequency amplitude in the wavelet domain. A function connecting the multi-coefficients, composed of arbitrary scaling on the lowest DWT coefficients, and the group SNR of these coefficient is derived. Karush-Kuhn-Tucker (KKT) theorem and minimum length play two essential roles to obtain the optimal modification of low-frequency amplitude with optimal scaling factors. Furthermore, the almost invariant feature of these optimized scaling factors hold demonstrates resistance to amplitude modification manipulation. To practically evaluate the watermarked audio quality, an objective measurement using the PEAQ is also performed as well. Experimental results confirm that the embedded audio in the proposed method has high SNR, low BER, and good PEAQ, indicating strong robustness against various attacks, such as re-sampling, amplitude modification, and mp3 compression.