In this study, we investigate in the use of hidden conditional random fields model to classify emotional speech. We introduce a novel hidden conditional random fields model, which is able to approximate complex distributions using a mixture of full covariance Gaussian density functions. In our experiments, we extracted Mel-frequency cepstral coefficients (MFCC) features from the well-known Berlin emotional speech dataset and eNTERFACE 2005 dataset. After that, we used the 10-fold cross validation rule to train, evaluate and compare our proposed model with the conventional learning method, hidden Markov model (HMM) and the existing hidden conditional random fields model, which can only utilize diagonal covariance Gaussian distributions. The experiments show that our method achieves significant improvement (p-value < 0.05) regarding the classification accuracy
Keyword
Emotion classification, Conditional Random Fields, HMM, GMM