一种基于动态分帧和掩蔽阈值分析的语音隐藏方法  

Speech Hiding Method Based on Dynamic Framing and Masking Threshold Analysis

作　　者：康凯 张林科 杨百龙 KANG Kai;ZHANG Linke;YANG Bailong(Rocket Force University of Engineering, Xi’an 710025, China)

机构地区：[1]火箭军工程大学,西安710025

出　　处：《兵器装备工程学报》2021年第9期159-165,共7页Journal of Ordnance Equipment Engineering

摘　　要：针对现有音频自适应隐藏算法存在掩蔽阈值精度和计算复杂度难以兼顾的问题,在实现动态分帧的前提下,通过对离散小波变换低频系数进行离散余弦变换实现了临界频带的准确划分,进而依据准确计算的掩蔽阈值自适应地选取嵌入强度,并采用容错处理和能量补偿方法消除了信息嵌入给嵌入强度盲检测带来的误差影响。仿真结果表明:所提方法在信息嵌入前后的语音听觉感知质量基本不变,隐藏容量较之对比方法提高1倍左右,且盲提取稳定可靠,运算复杂度未明显增加。In order to solve the problem to reconcile the accuracy of the masking threshold and the complexity of the calculation in the existing adaptive speech hiding method,a method based on dynamic framing and masking threshold analysis was proposed.Under the premise of carrier voice dynamic framing in the time-frequency domain,the critical frequency band was precisely divided by the discrete cosine transform for the low-frequency coefficients after a discrete wavelet transformation.And then an adaptive selection of embedding strength was achieved by accurately masking threshold calculation.Moreover,fault-tolerant processing and energy compensation was taken to eliminate the impact of the errors of embedding strength blind detection,which was caused by the information hiding process.The simulation results show that there was no significant decline in the quality of speech auditory perception before and after information embedding.And the proposed method has the advantages of no significantly increasing computational complexity but a stable and reliable blind extraction process,which could take 1 time higher hiding capacity than that of the contrast method.

关 键 词：自适应语音隐藏 掩蔽阈值分析 临界频带 能量补偿 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]