Perceptually transparent watermarking of audio signals

Digital watermarking captivates within remarkable promise as one of the silver bullets to fortify and insulate proprietary digital material in the times to come. It focuses on embedding watermark information into a digital host object in an inseparable bond to authenticate the latter. This paper proposes a high-capacity blind audio watermarking technique within perceptual constraints, i.e., the manipulations performed during the embedding process are imperceptible to unaided human senses. The suggested scheme uses high-frequency wavelet coefficients obtained during the quantization in Discrete Wavelet Transform (DWT) domain since redundant information can be accommodated with more ease in the high-frequency sub-band. Arnold Transformation is used for watermark permutation to ensure increased security and robustness of the scheme. From the experimental results, it follows that the illustrated scheme offers a healthy trade-off between payload, imperceptibility and robustness. The algorithm offers a fairly high Payload of 2755.6 bits per second while constantly maintaining the SNR, NCC and BER values well within the specified standards - even when been subjected to signal processing attacks.