A three-dimensional robust chaotic model and its application to dynamic cyclic dual-channel speech encryption

To address parameter discontinuities, accumulation of floating-point errors, and strong sample-to-sample correlation in existing speech-encryption schemes and to overcome the inability of single-channel models to exploit inter-channel correlation, this study proposes a dynamic cyclic dual-channel speech-encryption algorithm based on a three-dimensional exponential-sine robust chaotic mapping (3D-ESRC). The 3D-ESRC couples sine-based phase modulation with an exponential term to increase the sensitivity to initial conditions and suppress periodic degradation and numerical error accumulation, thereby expanding the effective key space. Building on this map, the proposed encryption framework uses composite nonlinear structures and dual-channel mappings with adaptive left–right coupling to realize the time-varying key-controlled scrambling of channel samples. A cyclic feedback mechanism implements hierarchical randomization across iterations, reducing the inter-sample correlation and increasing the confusion. The dual-channel architecture enables parallel processing and leverages dynamic coupling to strengthen the resistance against differential and classical attacks. Simulation and security analyses show that the proposed scheme achieves a lower post-encryption correlation, higher sensitivity to key and plaintext changes, and improved robustness compared with representative single-channel approaches. This method offers a practical and secure solution for dual-channel speech encryption, where both channel correlation and numerical stability are critical.