Abstract: In this paper，a new snore denoising method based on wavelet entropy is presented.According to snore signal characteristic and the theory of wavelet entropy，the serious disturbance and noise are removed and the snore signals are detected and extracted.Then clinical snore signal data is used for simulation，and different noise types with different SNR are analyzed.The results indicate that this method is simple，effective，especially for removing high-frequency noise and it is suitable for application.

Key words: snore, wavelet entropy, time-frequency distribution, denoising processing

摘要： 提出一种新的基于小波熵的弱鼾音信号去噪方法。结合鼾音信号特点及小波熵去噪理论，将严重的干扰和噪声去掉，并进行鼾音的检测及甄别，实现有效鼾音信号的提取，利用临床的实测鼾音信号数据进行模拟仿真，针对不同噪声类型和不同信噪比情况进行分析。结果表明，该方法简单有效，尤其对于高频噪声效果更优，且适于实际应用。

关键词: 鼾音, 小波熵, 时频分布, 去噪处理