Exploiting Psychoacoustic Properties to Achieve Transparent and Robust Audio Watermarking

This paper presents a novel scheme to adaptively determine the embedding strength for audio watermarking using quantization index modulation (QIM) in discrete wavelet packet transform (DWPT) domain. While the DWPT facilitates the signal analysis, the employment of a psychoacoustic model helps to not only acquire appropriate quantization steps for watermark embedding but also retrieve these steps back for watermark extraction. The effectiveness of this scheme has been proven using the perceptual evaluation of audio quality and bit error rates of the recovered watermark under common digital signal processing attacks. With the assistance of the psychoacoustic model, the combination of the DWPT and adaptive QIM attains satisfactory transparency and robustness with a payload as high as 183.03 bps.