Road Rage Recognition System Based on Speech Features

Abstract

A major cause of traffic accidents is road rage. How to identify road rage is an important problem that needs to be solved urgently. Road rage recognition is different from traditional emotion recognition. The sound signal to be recognized contains complex traffic environment noise, and the recognition target is a single anger emotion. This paper extracts high robustness, high generalization, and anger features from speech signals. A convolutional neural network (CNN) and multi-headed self-attention criterion bi-directional long-short-term memory network (Multi-headed Self-Attention Bi-LSTM) fusion decision model is proposed to realize anger emotion recognition.