Hiding data in audio files using coverless steganography methods

The paper proposes a coverless steganographic approach for hiding data in audio files. It is based on an architecture that includes the generative neural network model RealNVP for processing streaming data and auxiliary encoding and decoding modules. Piano-jazz was chosen as a musical genre for generating audio files. The results of experimental studies have confirmed the effectiveness of using streaming neural networks for the tasks of coverless steganography in audio files.