Emotional feature analysis and recognition in multilingual speech signal

In order to research the emotion recognition based on basic prosodic parameters, this paper discusses the statistical analysis of time, energy, pitch and formant parameters of multilingual emotional speeches, which includes Chinese, English, Russian, Korean and Japanese. A corpus of emotional speech spoken in multi-languages is collected. The statistical analysis shows that language factor doesn´t affect pitch variation features of some given emotion obviously. Principle Component Analysis (PCA) method is used to recognize the states of emotion in multilingual speech. According to the recognition results, we conclude that basic emotion states in multilingual speech can be recognized by a few simple prosodic parameters.