Based on EEMD-HHT Marginal Spectrum of Speech Emotion Recognition

Hilbert-Huang Transform is a time-frequency analysis method that apply to the nonlinear and non-stationary signal analysis, and has good adaptability. And The empirical mode decomposition(EMD) is the core part of HHT. Traditional EMD decomposition exists mode mixing phenomenon. To overcome this phenomenon, a new noise-assisted data analysis (NADA) method, the Ensemble EMD (EEMD), is proposed, which defines the true IMF components as the mean of an ensemble of trials and each consisting of the signal plus a white noise of finite amplitude. Finally the amplitude feature of emotional speech signal marginal spectrum is extracted using SVM classifier on emotional speech recognition.