Emotions analysis of speech for call classification

Most existing research in the area of emotions recognition has focused on short segments or utterances of speech. In this paper we propose a machine learning system for classifying the overall sentiment of long conversations as being Positive or Negative. Our system has three main phases, first it divides a call into short segments, second it applies machine learning to recognize the emotion for each segment, and finally it learns a binary classifier that takes the recognized emotions of individual segments as features. We investigate different approaches for this final phase by varying how emotions for individual segments are aggregated and also by varying classification model used for the final phase. We present our experimental results and analysis based on a simulated data set collected specifically for this research. © 2010 IEEE.