【摘要】 文章提出了一种基于Hilbert-Huang变换的汉语语音情感分类方法,研究了六种最基本的人类情感。用EMD分别对每一组的语音信号进行频率族的分解,并求其Hilbert能量谱,用以反映情感语音信号的能量随时间和频率的分布情况,从而分离出不同的情感。结果表明,基于Hilbert-Huang变换可以非常有效的识别出各情感语句中所带的情感。Hilbert-Huang变换作为一种新的信号分析方法,被成功的应用到了语音情感识别领域中来,这为语音信号情感分析提供了一个新的可行的研究方法,也为拓展Hilbert-Huang变换的理论应用进行了一个新的尝试。更多还原

【Abstract】 A speaker emotion identification method based on Hilbert-Huang Transform(HHT)is proposed in this paper,six basic human emotions are investigated. Intrinsic mode functions are obtained by empirical mode decomposition(EMD)on signal,the EMD allows us to study the different intrinsic oscillatory modes and instantaneous frequencies of the trace.It provides time-frequency attributes and can separate different emotions.The experiment results indicate such method works well in speaker emotion identification.It is the first time attempt to apply the theory of HHT to detect emotion from a speech signal.更多还原