Design and Implementation of the Teacher-student Dialogue Automatic Speech Recognition Tool Based on Classroom Verbal Characteristics
Gang Lian Zhao, Jiarong Yi, Jie Chu, Yinan Zhang, Jianghua Yin · 2022
Speech behavior is a tool for teachers and students to express and communicate in the classroom, and analyzing the content of classroom teacher-student dialogue is one of the key basic technologies to break through the automatic analysis of teaching videos. However, the current discourse research tools rely on the universal speech recognition cloud platform, and there are problems such as confusion between teacher-student role and unclear discourse boundaries for classroom teacher-student dialogue recognition. Therefore, this paper designs and develops teacher-student dialogue recognition tools after classifing and encoding the auditory information data and summarizing the general rules of speech characteristics in classroom teaching videos. To achieve the purpose of analyzing classroom teaching videos, education researchers can use the tool to intercept teaching videos, obtain boundary points of teacher-student dialogue, and recognize their speech content. Experimental results show that the tool has stable performance, which can quickly and accurately process audio signals, segment and cluster teacher and student voiceprint features. Finally, it produce structured text containing the time point of the discourse boundary, the teacher-student role label and the discourse content, which will provide data support for analyzing further classroom teaching video behavior.