Real-time Assessment of Student Concentration Based on Attention and Speech Emotion Recognition

Abstract

Nowadays, education is mostly in the form of big-scale class education, which will cause that teachers cannot pay attention to all students in real time. Therefore, how to assist teachers to detect attention of students in real time has become a problem that we need to consider. We propose a method with mechanism of attention for real-time assessment of student concentration based on speech emotion recognition. By comparing different classification algorithms, CNN is selected to process speech to generate emotion classification probability. This paper introduces new learning parameters and puts forward a formula to obtain the attention score to monitor the student concentration status better through speech emotion. The research provides a new direction for the application of deep learning in auxiliary education. This project will help both teachers and learners to improve their way of teaching and learning through calculated concentration score.