Robust, transparent and high-capacity audio watermarking in DCT domain

摘要

This paper presents a novel scheme capable of achieving robust, transparent and high-capacity blind audio watermarking. The watermark embedding is performed by modulating the vectors in the DCT domain subject to an auditory masking constraint. An algorithm has been developed to maintain the energy balance in a frequency band, thus allowing the retrieval of embedded information directly from the designated band. As the proposed watermarking scheme is implemented on a frame basis, the abrupt artefacts in frame boundaries are further rectified via linear interpolation over transition areas. The embedded watermark is shown to be imperceptible to human ears and yet robust against common digital signal processing attacks. The resulting payload capacity is as high as 848 bps. Moreover, not only a 100% recovery of the watermark is guaranteed for non-attack situations but the survival rate is substantially improved in the case of lowpass filtering even at a cutoff frequency as low as 500 Hz.