Double Semantic Watermark Algorithm for Digital Audio Based on Neural Network

The researches on digital watermarking algorithm based on semantics are hot issues in the field of audio content management. The indexing method of semantic management for audio watermarking based on uniform content locator (UCL) is introduced, and a dual watermarking algorithm based on radical basis function neural network (RBFNN) is proposed. In semantic watermarking model, RBFNN is used to adaptively select the best embedding place of watermarking in the audio data segment, and the wavelet transform is also adopted to extract the approximate weight and the detail component of the selected audio segment. The different audio indexing information as dual semantic watermarking are embedded into the corresponding audio signal, and the constructed dual semantic watermarking model can realize the integration transmission target with semantic information and original audio signal. Furthermore, synchronous code technology is utilized to solve the problem of effective watermarking detection and monitoring. The results show that the proposed strategy can bring excellent robustness and non-auditory while the embedded watermarking is semantic information with large amount of information.