Robust audio watermarking based on multi-carrier modulation

摘要

The great challenge, in blind audio watermarking, is to recover the transparent embedded information with minimum errors even in presence of disturbances. Some signal disturbances like MPEG compression and filtering can be modelled as fading-like distortions. Besides the inaudibility constraint, the problem will be viewed as the conception of a particular communication system, where our purpose is to reduce host interference noise (HIN) and inter-symbol interference (ISI) on one hand and to deal with fading perturbations on the other hand. In this paper, we propose adapting multi-carrier (MC) modulation, widely used in wireless communications, to the watermarking context. Our main motivation is to exploit advantages of MC modulations in order to outperform spread spectrum (SS) audio watermarking schemes. The new MC-watermarking system allows us to improve detection reliability and to increase robustness of the embedded watermark. Moreover, the inaudibility constraint will be ensured by adapting the MC signal parameters to the masking characteristics of the human auditory system (HAS).