Identification of Key Parameters for Cover Audio Generation in Imperceptible Audio Steganography

The security of data during communication is an ever-growing challenge, as outdated algorithms face increasing vulnerability. Audio steganography offers a promising alternative by embedding data within audio signals in a manner that is imperceptible to human listeners. Recent advancements in cover selection as well as cover generation techniques have improved the quality and suitability of audio for embedding, yet the parameters influencing imperceptibility remain underexplored. This research aims to identify and evaluate key audio parameters which are energy, the total and value distribution of silent samples. These parameters can be assessed prior to the embedding process which offer a potential advantage over existing metrics, typically require post-embedding analysis to accurately assess imperceptibility. The methodology of this research consists of five main stages: parameter identification, dataset preparation, parameter extraction, embedding, and evaluation. Signal-to-Noise Ratio (SNR) was employed as the primary metric to assess perceptual transparency, with 20% data capacity embedded in each audio of total 220 audio. The first experiment revealed a positive correlation as high as r = 0.706 between energy and SNR, indicating that higher-energy audio benefit from auditory masking, resulting in improved imperceptibility. The second experiment showed a negative correlation, r = –0.787 between total silent sample and SNR, confirming that embedding in silent sample increases the risk of perceptible artifacts. Silent sample that has higher value exhibited superior SNR values compared to those with extremely lower value. These findings highlight the importance of pre-embedding analysis in identifying suitable samples for embedding. By identifying and evaluating these influential parameters, this research contributes to the development of intelligent cover generation strategies that enhance the transparency of audio steganography systems. The insights presented offer a foundation for future design, optimization, and real-world application of robust data-hiding techniques.