A Generalized Spread Spectrum-Based Audio Watermarking Method with Adaptive Synchronization

In this paper, a new audio watermarking scheme is proposed that addresses the synchronization problem using an adaptive filter. To spread the watermark energy across the spectrum of the host audio signal, the scheme uses Hamming coding, convolutional encoding, and generalized PN sequence generation and for watermark recovery, the scheme utilizes Viterbi decoding. Also, the watermark cannot be detected either statistically or perceptually. To increase the watermark embedding capacity, the watermark is embedded in certain areas of the host signal, which increases the masking threshold. By using a high masking threshold, the adaptive watermark insertion, blind detection, robustness, and capacity of watermarking are enhanced in comparison to similar methods. Finally, experimental results show that in addition to guaranteeing an error bit rate (BER) of almost 0 for non- and minor-attack conditions, the BER is substantially improved in the case of low-pass filtering at a cutoff frequency as low as 3.2 KHz.