This paper proposes a novel approach of audio watermarking based on Spread Spectrum (SS) involves the psychoacoustic model and deep learning Convolutional Neural Networks (CNN)-autoencoder. Moreover, logistic chaotic maps are employed to enhance the security level of the method. First, a compressed image produced from the CNN-autoencoder is fed to the image encryption stage to yield an encrypted image to be used as the watermark. To apply image encryption, the plain image is, at first 8-bit binary-coded and shuffled by M-sequence. Next, each encoded image is diffused with a different chaotic sequence. Within the embedding phase, the psychoacoustic model is employed to shape the amplitude of the watermark signal which guarantees high inaudibility, whereas a logistic chaotic map is used to determine the positions for watermark embedding in a random manner. This scheme offers an extremely efficient and practical method as it can be used by institutions and companies for embedding their logos or trademarks as a watermark in audio products as the scheme utilizes RGB images. Experimental results show that the transparency and imperceptibility of the proposed algorithm are satisfactory also good image quality even against various attacks. The validity of the proposed audio watermarking method is demonstrated by simulation results.
(See document for exact affiliation information.)
https://aes2.org/publications/elibrary-page/?id=21584
(418KB)
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member Join the AES. If you need to check your member status, login to the Member Portal.
