Abstract
In our work, we propose an improved voice encryption algorithm based on chaotic dynamics that makes the voice encryption algorithm more secure against attacks. The unique contribution of this work is the combination of SHA3-512 hash function, Empirical Mode Decomposition (EMD) operations, and an optimized random bit pool in a high-integrity encryption framework. The distinguishing feature is that the encryption key is obtained from a cryptographic quality random bit pool generated by optimization-based generators, exhibiting good statistical properties. This structure offers unpredictability beyond known key generation mechanisms. At the same time, the signal adaptive feature of EMD, encrypting only the residue component, has increased the computational efficiency and improved the algorithm's suitability for real-time applications. The dependence of the chaotic parameters on the seed value conditions and the use of a hybrid with an optimized random bit pool significantly improves the reliability of the encryption system. Experimental analysis statistically demonstrates that the algorithm shows low correlation, high entropy growth, and satisfactory decryption accuracy. Thus, it proves to be a significant contribution to the literature in terms of both cryptographic security and computational efficiency for voice encryption operations.