Robust audio watermarking technique based on gammatone filterbank and coded-image

The essence of watermarking technique is imperceptibly watermarking the audio media with a secret watermark for copyright protection. Robustness is one vital property of any practical watermarking scheme. To claim its effect on ownership verification, a watermarking scheme must withstand various dasiaattackspsila, including ordinary signal processing operations and malicious tampering. In this paper, we present a robust audio watermarking system based on gammatone auditory filterbank (GTF) and coded-image. Instead of uniform filterbank, gammatone filterbank is utilized to obtain non-linear subbands that resemble the frequency response of the cochlea. gammatone filterbank has wider bandwidths at higher frequencies and narrower bandwidths at lower frequencies, which contributes to achieving transparent perception in the process of embedding the watermark without degradation on robustness. Moreover, we embed coded-image as a kind of visual watermark, other than pseudorandom sequence. Simulation results show that our scheme exhibits remarkable resistance against common manipulations and deliberate attacks, such as adding noise, low-pass filtering, resampling, lossy compression, random samples cropping and jittering.