Audio steganography using discrete wavelet transform (DWT)

The rapid growth of digital communication has increased the risk of unauthorized access to sensitive information. While traditional cryptographic techniques ensure data confidentiality, they do not conceal the existence of communication and may therefore raise suspicion in certain contexts. Audio steganography addresses this limitation by embedding secret information into audio signals; however, many existing approaches face inherent trade-offs between transparency, robustness, and capacity. This thesis presents a hybrid audio steganography framework that combines Discrete Wavelet Transform (ADD) and Quantization Index Modulation (NİM) for transform-domain embedding, together with GŞS-ŞBZ encryption to protect message confidentiality prior to insertion. Encrypted textual data are embedded in high-frequency wavelet coefficients of uncompressed WAV audio, exploiting properties of the human auditory system to preserve perceptual quality. A prototype implementation was developed in Python and evaluated using Peak Signal-to-Noise Ratio (PSNR) and Bit Error Rate (BER). The experiments deliberately focused on controlled, uncompressed audio conditions in order to isolate algorithmic behaviour. The results indicate very high perceptual transparency but also reveal persistent bit errors at the decoder, particularly when encryption is applied. These findings indicate that message recoverability is partial and constrained by numerical effects in the transform and reconstruction stages, highlighting a structural tension between perceptual inaudibility and bit-level recoverability rather than reliable extraction. Overall, the work provides an interpreTablo baseline framework that integrates transform-domain steganography with modern cryptography. The system is primarily suited for controlled or offline scenarios where transparency is prioritised, and it serves as a foundation for future research targeting robustness-oriented enhancements and broader real-world evaluation.