FGAS: Fixed Decoder Network-Based Audio Steganography with Adversarial Perturbation Generation

Abstract: The rapid development of Artificial Intelligence Generated Content (AIGC) has made high-fidelity generated audio widely available across the Internet, providing diverse cover signals for covert communication. Driven by advances in deep learning, current audio steganography schemes are mainly based on encoding-decoding network architectures. While these methods greatly improve the security of audio steganography, they typically require complex training and large pre-trained models. To address the aforementioned issues, this paper pioneers a Fixed Decoder Network-Based Audio Steganography with Adversarial Perturbation Generation (FGAS). Adversarial perturbations carrying secret message are embedded into the cover audio to generate stego audio. The receiver only needs to share the structure and weights of the fixed decoder network to accurately extract the secret message from the stego audio, this eliminates the reliance on large pre-trained models. In FGAS, we propose an audio Adversarial Perturbation Generation (APG) strategy and design a lightweight fixed decoder. The fixed decoder guarantees reliable extraction of the hidden message, while the adversarial perturbations are optimized to keep the stego audio perceptually and statistically close to the cover audio, thereby improving resistance to steganalysis. The experimental results show that FGAS significantly improves the quality of stego audio, achieving an average PSNR gain of over 10 dB compared to SOTA methods. Moreover, FGAS exhibits superior anti-steganalysis performance under different relative payloads; under high-capacity embedding, it achieves a classification error rate about 2% higher, indicating stronger anti-steganalysis performance compared to current SOTA methods.