基于语音参数模型的语音隐藏算法

基于语音参数模型，该文提出一种将保密语音隐藏在公开语音中的信息隐藏算法．首先将保密语音经混合激励线性预测(MELP)编码和纠错编码形成隐藏信息．然后根据瞬态互相关基音周期检测算法确定频域嵌入点，并通过修改对应的DFT系数来隐藏信息．提取时按同样方法确定嵌入点恢复隐藏信息，并经MELP解码还原保密语音．实验结果表明嵌入信息后，中间语音的分段平均信噪比接近60dB，并且在受到压缩、滤波等攻击时具有较强的鲁棒性．算法为信息安全和数字水印领域研究开辟了一条新的途径．

Speech Hiding Algorithm Based on Speech Parameter Model

Based on speech parameter model, a novel approach to embed secret speech into open speech is proposed in this paper. First, secret speech is coded into binary parameter bits with Mix-Excitation Linear Prediction (MELP) algorithm, and the bits are arranged to form hiding information sequence. Then open speech is divided into several frames for information embedding. On every frame, according to crosscorrelation-based pitch detection method, the instantaneous pitch is utilized to determine the current embedding position in frequency domain of open speech. Finally, information hiding procedure is completed by modifying relevant frequency coefficients, which is controlled with current information bit. At the receiver, based on the same pitch detection method, the embedding position is found and the hiding bit is recovered sequentially. After MELP decoding, the secret speech waveform can be recovered entirely. The experimental results show that the average SNR of middle speech is close to 60dB, and the watermarking is strongly robust to many attacks such as compression, filter and so on. This algorithm provides a new approach for the research on information security and digital watermarking.