Robust and high-capacity audio watermarking basedon chirp z-transform

Abstract

A novel non-blind audio watermarking algorithm, using chirp z-transform in combination with discrete wavelet transform and singular value decomposition, is presented in this paper. In the proposed method, a watermark audio signal (about 13 seconds in length) is hidden into an original audio signal (about 51 seconds in length) by applying discrete wavelet transform, chirp z-transform, and singular value decomposition. The effectiveness of the proposed algorithm has been tested using signal-to-noise ratio, objective difference grade (ODG), and normalised cross- correlation of watermarked audio signal under various common audio attacks. Experimental results demonstrate that the algorithm has a high capacity (352.8 kbps), without significant perceptual distortion (ODG > -0.29) and provides good robustness against common audio attacks, including: re-sampling, echo addition, MP3 compression, additive white Gaussian noise, re- quantisation, cutting, and low-pass filtering. This algorithm has a high capacity and excellent audio quality compared with other audio watermarking algorithms.