Nonlinear Dynamic Feature Extraction Based on Phase Space Reconstruction for the Classification of Speech and Emotion

<div>Principal component-spectrum analysis of different emotional speeches from CASIA emotional speech corpus. Principal component-spectrum analysis of (a) happiness, (b) sadness, (c) neutral, and (d) anger.</div>

Mathematical Problems in Engineering

fig2

Figure 2

Figure 2: Nonlinear Dynamic Feature Extraction Based on Phase Space Reconstruction for the Classification of Speech and Emotion 