Advanced Audio Coding (AAC) is currently the most widely used audio compression standard in Internet applications. In this paper, a large capacity secure steganography algorithm based on the compression parameters of AAC QMDCT is proposed. In order to improve the statistical security and the imperceptibility of steganography. Based on multi-view statistical distortion, which combines the statistical distribution of Huffman codewords, QMDCT’s intra-frame and inter-frame statistical correlation, and the frequency perceptual masking in the psychoacoustic model, this paper presents a secure AAC steganography scheme SofMvD based on STCs. In the experimental results, the maximum embedding capacity of the proposed scheme is about 12 kbps. Compared with the existing methods, this method can improve steganalysis’s ability to $8\text{\%}-30\text{\%}$ and improve the auditory concealment of steganalysis without changing the embedding capacity. This scheme can also be extended to MP3.

高级音频编码（AAC）当前是Internet应用程序中使用最广泛的音频压缩标准。提出了一种基于AAC QMDCT压缩参数的大容量安全隐写算法。为了提高统计安全性和隐写术的隐蔽性。基于多视角统计失真，结合了霍夫曼码字的统计分布，QMDCT的帧内和帧间统计相关性以及心理声学模型中的频率感知掩蔽，提出了一种安全的AAC隐写方案SofMvD基于STC。在实验结果中，该方案的最大嵌入容量约为12 kbps。与现有方法相比，该方法可以提高隐写分析的能力。$8\text{％}-30\text{％}$并在不改变嵌入能力的情况下改善隐写分析的听觉遮盖力。该方案也可以扩展到MP3。