Multiple Audio Watermarking With Spectrum Pyramid

The paper presents a new approach for intellectual property protection of audio products when they are distributed via computer networks or radio broadcasting. The presented approach is based on a method for digital watermarking, performed in the spectrum domain, using a Complex Hadamard Transform (CHT) with a specific matrix, in which only 1⁄4 of the elements are complex numbers. In correspondence with this method the watermark data modifies the phases of selected spectrum coefficients of the audio signal. The values of these coefficients are used to build the Spectral Pyramid (SP), which is the synonymous presentation of the processed audio signal. The pyramid permits multi-layer watermarking, because the watermarks, embedded in its layers do not interchange. The paper presents the algorithms for the audio watermark embedding, extraction and detection. The method is evaluated on the basis of some theoretical analysis comparisons and on experimental results. The new approach presents applications in e-commerce, distance learning, tracing and proving of the illegal distribution of audio products.