摘要
提出了一种基于EMD的功率谱分析法,即先用经验模态分解方法(EMD)将语音信号分解成若干个固有模态函数IMF分量,而后对包含主要信息的IMF分量利用现代参数模型法估计出它们各自的功率谱。文中对不同情感状态下语音数据的分析结果表明,EMD方法能有效地应用于非平稳语音信号的功率谱分析中,可更细致的体现语音信号内在特征。
A new power spectrum analysis of speech signal method based on Empirical Mode Decomposition is presented. First, the speech signal is decomposed into several intrinsic mode functions （IMF） by using empirical mode decomposition （EMD） method. Then, power spectrums of some IMF components which contain main information calculated by modem PSD estimation method. The analysis results of the experimental emotion speech data show that the proposed approach can be applied to the power spectrum analysis of non-stationary speech signal efficiently, can be more detailed inherent characteristics of the speech signal.
出处
《电子设计工程》
2012年第4期162-164,共3页
Electronic Design Engineering
基金
国家科技支撑计划(2006BAD15B06)
