MODELS AND ALGORITHMS FOR CREATING DIGITAL WATERMARKS FOR AUDIO FILES

Abstract

To justify and investigate models and algorithms for creating digital watermarks in audio files based on the construction of neural network compressive maps, in the interests of increasing stealth and stability, as well as ensuring the universality of algorithms for creating CVZ. The methods and approaches of the modern theory of information processes and systems were used, namely: methods of the theory of statistical solutions; methods of probability theory and mathematical statistics; technologies of object-oriented programming; theory of random processes and fields, methods and technologies of statistical simulation modeling; methods of digital processing and recognition of signals and images; apparatus of artificial neural networks; methods of computer steganography, as well as the technology of direct computer experiment, which is performed in relation to real objects of digital content. Algorithms for creating digital watermarks for audio format objects have been obtained. Algorithms for creating digital watermarks for objects, the format of data presentation, as well as researching their possibility for the use of sound format files. Methods and results of the analysis of statistical visibility and the possibility of restoring the built-in sequence of TVS by a third-party observer for assessing the quality of TVS in objects of sound formats. The expediency of building and using universal compressive transformations for steganographic embedding of CVZ into container objects of various types with a minimum level of dispersion of distortions on the basis of artificial two-layer forward propagation neural networks has been established. The results of the research are of practical importance for the development of special mathematical and software in the interests of creating digital watermarks as means of hidden marking of digital content objects, which provide effective control of the use of intellectual property objects, as well as for audio data diagnostics.