Speech emotion recognition based on a modified brain emotional learning model

This paper introduces an optimized model of brain emotional learning (BEL) that merges the Adaptive Neuro-Fuzzy Inference System (ANFIS) and Multilayer Perceptron (MLP) for speech emotion recognition. This model is developed based on the limbic system of the mammalian brain in order to present a desirable learning model for speech emotion recognition in dynamic situations like the brain's emotional networks. The proposed model has four main parts including thalamus, sensory cortex, orbitofrontal cortex and amygdala. As this model is inspired from human brain, ANFIS has been used in this study to develop a model in subsystems of amygdala and orbitofrontal cortex to make rules. To classify the rate of speech emotion signals, the ANFIS outputs are given to MLP network. Experimental results show that the proposed algorithm has a powerful capability to identify a human's speech emotion.