Research And Application On Mobile Speech Emotion Recognition System

As we all know, language is a useful tool in communication. There isa plenty of language information as well as emotional information inhuman being’s language. And the speech emotion recognition plays a vitalrole in the field of signal processing and artificial intelligence. The currentteaching environment, especially with the widely use of E-learning[1]andM-Learning[2], leads to a lack of emotional communication betweenteachers and students. While the emotional feedback, as a kind ofimportant feedback information, can help teachers to adjust the teachingstrategy according to students’ emotion and improve the teaching equality.This article studied and developed a speech emotion recognitionsystem based on the mobile devices, which can analyze the speaker’sspeech signal in real time, recognize and record the current emotion states,and do the post processing and statistics on the collected data.Nowadays as the technology of speech emotion recognition is still atthe beginning of the stage, no standard algorithm is widely used for thefeature selection and extraction. After doing research and experiment onthe algorithm, we improved the current algorithm on feature selection andextraction, and used SVM (Support Vector Machine) to obtain a modulewith high ability of emotion recognition. The performance on Mobiledevice is limited compared with PC. This article trained the data on PC, sothat mobile devices can make use of the module we get from PC directly todo emotion recognition, which also improves the current processingefficiency. In addition, this article takes advantage of the recognitionsystem to collect emotion data and do statistics with the combination of thedaily activities happens during the period with the emotion data to verify the emotion changes.