Speech Emotion Recognition Using MFCCs Extracted from a Mobile Terminal based on ETSI Front End

摘要

The importance of automatically recognizing emotions from human speech has grown with the increasing role of spoken language interface in man-machine applications. The paper presents a system for the recognition of emotional states based on parameters extracted at the front end of a mobile terminal according to the ETSI ES 202 050 standard. Starting from a vector of various features derived from energy and MFCCs, an approach based on genetic algorithms is used to determine a subset of features that will allow robust speech classification of 7 emotional states: anger, joy, sadness, fear, disgust, boredom and neutral.