基于经验模态分解和Teager峭度的语音端点检测

Endpoint detection of speech signal based on empirical mode decomposition and Teager kurtosis

采用经验模态分解和Teager峭度的统计特性对噪声环境下的语音信号端点进行检测。利用经验模态分解获得语音信号的本征模态函数,用Teager能量算子计算每个本征模态函数的瞬时能量,并对本征模态函数进行系数—峭度计算,提取信号期望的统计特征信息实现语音端点的检测。通过自适应EMD分解和Teager能量算子的处理,这种方法可以有效地消除白噪声或有色高斯噪声的影响。通过仿真例子说明这种方法可以取得良好的端点检测效果,仿真研究结果表明用经验模态分解和Teager峭度对噪声环境下的语音端点检测是可行的和有效的,提高了检测的可靠性。

A new algorithm for endpoint detection of speech signals in noisy environments based on empirical mode decomposition（EMD） and statistical properties of Teager kurtosis is proposed.The speech signal is firstly decomposed into intrinsic mode function（IMF） using empirical mode decomposition method.Then Teager energy operator is used to track the modulation energy of each IMF.The desired feature of statistical properties of speech signals can be extracted from the coefficient-kurtosis value of the intrinsic mode function.Through self-adaptive decomposition with EMD and Teager energy operator processing,the proposed method can effectively eliminate the disturbance of additive white or colored Gaussian noises.In order to show the effectiveness of the proposed method,we present examples showing that the new method is more effective than traditional methods.Experiment results show the feasibility and efficiency of the EMD and Teager kurtosis method in endpoint detection of speech signals in noisy environment; additionally,the algorithm is very reliable to be implemented for endpoint detection.