Affective Model Based Speech Emotion Recognition Using Deep Learning Techniques

Abstract

Human beings express emotions in multiple ways. Some common ways that emotions are expressed are through writing, speech, facial expression, body language or gesture. In general, it is believed that emotions are, first and foremost, internal feelings and experience. Speech is a powerful form of communication that is accompanied by the speaker's emotions. Specific prosodic signs, such as pitch variation, frequency, speech speed, rhythm, and voice quality, are accessible to speakers to express and listeners to interpret and decode the full spoken message. This paper aims to establish an affective model based speech emotion recognition system using deep learning techniques such as RNN with LSTM on German and English Language datasets.