Reduce the dimensions of emotional features by principal component analysis for speech emotion recognition

In this paper, the principal component analysis (PCA) is applied to speech emotion recognition for improving the accuracy of the system. The traditional prosodic features like pitch-related features and formant-related features are extracted from the Berlin speech database [7] and a Chinese database. These collected feature data is processed by PCA to remove the irrelevant information. After that, three kinds of features including the processed features by PCA, unprocessed features and other speech-related features are used to train a SVM classifier. And six emotions are tested in the experiment. The classification accuracy of the processed features by PCA is about 3.1% higher than the unprocessed features and about 17.6% higher than the MFCC features when using 240 utterances. The recognition accuracies among different emotions in both databases are also presented in the study.