Secure Audio Steganography using Vectorized LSB and Chaos-Based Encryption

In the era of digital transformation, the protection of sensitive multimedia data against growing cyber threats has become increasingly critical. Traditional cryptographic and steganographic techniques, while effective individually, often fall short when faced with advanced detection and attack methods, making hybrid security approaches a necessity. In this study, a hybrid security approach combining chaotic algorithms and the Least Significant Bit (LSB) embedding method is proposed. The method is enhanced through key-dependent parameter assignment, as well as additional steps such as transient periods and square matrix transformation. Furthermore, vectorization after square matrix transformation simplified indexing in embedding and extraction steps, thereby improving computational efficiency. As a result, high security and integrity were achieved for both visual and audio data. In the encryption process, seven different chaotic structures (Logistic Map, Lorenz System, Piecewise Linear Chaotic Map, Tent Map, Hénon Map, Chua Circuit, Chebyshev Map) were supported, and the method was tested on these algorithms. The original content was encrypted using XOR and then embedded into the audio signal via the LSB method. The proposed method was evaluated using the EBU SQAM audio dataset and standard test images with Peak Signal-to-Noise Ratio (PSNR), Signal-to-Noise Ratio (SNR), Mean Squared Error (MSE), Structural Similarity Index Measure (SSIM), and Bit Error Rate (BER) metrics. Experimental results demonstrated that for all tests, the extracted images achieved Visual_PSNR = ∞, Visual_SSIM = 1, and Visual_BER = 0. For audio data, PSNR values ranged between 102–107 dB and SNR values approximately 53–99 dB depending on the content type. These findings reveal that the proposed method ensures both the integrity of the content and the reliable preservation of the hidden data.