骨导气导结合的语音增强算法

Speech Enhancement Algorithm by Combining Bone-Air Conduction Signals

通过对骨导气导信号的对比和研究,提出了一种骨导气导结合的语音增强算法。它充分利用了骨导信号屏蔽环境噪声的高信噪比特性,将骨导信息以概率模式引入到改进后的维纳滤波算法中,并针对高噪声环境下语音信息被噪声严重破坏的情况,将骨导信号通过能量对比得到的权系数修正替换到原气导信号,实现了较好的语音增强效果。

A speech enhancement algorithm with combined bone-air conduction features is proposed in this paper, by leading a research on bone-conduction signals and air-conduction signals. According to the high signal noise ratio of bone-conduction signal, the probability of speech present in bone-conduction signal is obtained to improve the gain function and noise estimation of Wiener algorithm. In a rigorously noisy environment a weighted coefficient is employed to convert bone-conduction signal to low frequency air-conduction signal. The experimental results confirm the validity of the proposed method.