Robust zero-watermarking algorithm based on audio beats

Abstract

Beat can be used to describe the integrity of audio. In order to achieve audio integrity authentication more effectively, a robust zero-watermarking algorithm based on music beat extraction is proposed. Firstly, the beat of music audio is extracted after preprocessing, and then the audio are dynamically divided into segments according to the beat. After discrete wavelet transformation (DWT) and single value decomposition (SVD), the maximum singular values are encoded based on K-means algorithm. Since the maximum singular value is robust to various attacks, the information encoded according to its value is selected as audio feature information. After the original watermark and the feature information is XOR, zero-watermark is generated by voting mechanism. The experimental results show that the proposed algorithm is robust against common attacks. In addition, the algorithm can also resist many synchronization attacks such as TSM and jittering.