A novel audio steganography based on the segmentation of the foreground and background of audio
Jinglei Wang, Kaixi Wang · Computers & Electrical Engineering · 2025
There is a growing concern regarding data transmission security in public network. Steganography plays a crucial role in transferring data in a secure and reliable way in the environment. Audio steganography is a technique that embeds a secret into audio without arousing any suspicion from hearing it. A novel audio steganography is proposed based on the segmentation of the foreground and background of digital audio in this paper. Due to the distinct sensitivities of human ear to the foreground and background of audio, and a modification in the foreground areas has less impact on the quality of audio than that in the background areas, thus different areas of audio should be embedded in a different embedding ratio. To improve the embedding capacity, an adaptive threshold is computed for a given cover to segment it into the foreground and background areas, thus more bits are embedded into the sampling points in the foreground areas than in the background areas. Also, a random selection algorithm is employed to choose sampling points to embed to enhance security. The experimental results show that the proposed method has a higher embedding capacity and less impact on its imperceptibility and security than Psy_steganography, and its average maximum embedding capacity reaches 24,574 bits calculated by segmenting one hundred audios from TIMIT dataset with their adaptive thresholds, and it is 53.59% higher than Psy_steganography.