一种低速率压缩语音通用隐写分析方法

A Common Steganalysis Method for Steganography in Low-bit-rate Speech

合成分析线性预测编码(AbS-LPC)广泛应用于多种低速率语音编码中,现有的AbS-LPC低速率压缩语音隐写检测方法针对特定类型的隐写方法而设计,泛化能力较弱。为此,本文提出了一种面向多类低速率压缩语音隐写的通用信息隐藏检测方法。由于AbS-LPC低速率压缩语音码流中的码元存在时空关联性,且所有AbS-LPC低速率压缩语音隐写方法本质上都是改变码元取值。因此,从码元的角度出发,基于贝叶斯推理实现多类低速率压缩语音隐写的通用隐写检测。实验结果证明本文方法具有很好的通用隐写检测效果。

Analysis by Synthesis-Linear Predictive Coding(AbS-LPC) is widely used in many low-bit-rate speech codecs. The existing steganalysis methods are lack of generalization ability, because they are designed for a specific type of AbS-LPC low-bit-rate compressed speech steganography. In this paper, a common steganalysis method for steganography in low-bit-rate speech is proposed. The code elements in AbS-LPC low-bit-rate compressed speech steam exist spatiotemporal correlation, and the steganography will ultimately change the values of these code elements. Therefore, from the perspective of code elements, this paper proposes a common steganalysis method based on Bayesian inference. The experimental results demonstrate that the proposed method can achieve good performance.