Balancing the Trade-Off Between Capacity and Imperceptibility for Least Significant Bit Audio Steganography Method: A New Parameter
Muhammad Harith Noor Azam, Farida Hazwani Mohd Ridzuan, M. Norazizi Sham Mohd Sayuti, Ahmed A. AlSabhany · 2019
Data is vulnerable to unauthorized interception communication hence steganography is used to protect it. However, there is a trade-off between capacity and imperceptibility features in audio steganography. Based on the existing study, number of bit embedded per sample and the ratio of size secret message to the number of cover audio sample affects the quality of stego-file but lack of focus was given on reducing the trade-off problem. An evaluation study on 6 Least Significant Bit (LSB) embedding methods was carried out to capture the performance of the stego-file based on Peak Signal to Noise Ratio (PSNR). The selected cover audio files were 44.1 kHz mono with 16-bit depth. Different size of data was embedded at 30 different initial embedding points. Based on that study, this paper proposes a new parameter to improve the imperceptibility feature while maintaining the capacity feature. A new parameter which is the initial embedding point is proposed to maximize imperceptibility given a capacity value. The experimental result shows that there is an effect by selecting the initial embedding point towards imperceptibility features of each LSB method. However, the significant effect was only recorded by the embedding method at the 4th level and higher levels of LSB. In conclusion, the selection on the initial embedding point could improve the imperceptibility feature while maintaining the capacity feature at the higher levels of LSB embedding. However, the selection on initial embedding point will become completely insignificant if bigger size of data is embedded and become completely insignificant at 0.05 standard deviation, as the embedded data increases.