Analysis Of Classroom Interactive Behavior Based On Voiceprint Recognition

In recent years,artificial intelligence technology has begun to integrate into various fields,and has brought about huge changes,especially the education industry.The traditional classroom is still the most important learning environment in the current education.How to evaluate the teaching process in the classroom is still a difficult problem.While the development of technology has brought unlimited possibilities to classroom performance evaluation.This paper utilizes voiceprint recognition technology to analyze and evaluate classroom interactions,determine classroom interaction modes and classroom structure.The study was conducted in following parts:Firstly,the article introduced the research background of classroom analysis.And the research status of voiceprint recognition and classroom interaction behavior analysis at home and abroad were summarized and analyzed.At the same time,the paper explained the theories of voiceprint recognition,classroom interaction analysis and social network analysis used in classroom interaction analysis based on voiceprint recognition.Then,the paper introduced the process of speaker diarization(speaker recognition and classification).First,data collection and feature extraction are performed.Then,conduct active voice detection on audio data.Further,speaker changes is detected.Finally,perform the classification to group the same speaker utterance fragments.Then it introduces the process of speaker recognition and clustering.After obtaining the results of voiceprint recognition and classification,classroom interaction analysis can be performed.Finally,cases analysis were made on the above-mentioned classroom interaction behavior analysis method based on voiceprint recognition,and the actual cases were concretely calculated and analyzed.Starting from single-case and multiple-case comparisons,voiceprint recognition and classification tuning parameters were performed in single-case analysis,and interactive data comparisons of multiple courses were conducted in multiple-case comparison analysis.Finally,the feasibility and scientificity of classroom interaction analysis based on voiceprint recognition were confirmed.Meanwhile,the research content and existing problems were summarized.The main results of the research are:(1)Speaker diarization are used in classroom data,and classroom audio is processed to achieve automatic classroom analysis,solving the problem of inaccurate classroom interaction analysis.(2)The automatic analysis of ST analysis is realized,which does not need to manually mark the behavior in each observation time,and also consumes little manpower to analyze a large amount of data,avoiding the time-consuming and laborious manual analysis and subjective judgment mistake.(3)Use social network analysis for offline classrooms,draw a social network graph that visualizes the classroom,and more intuitively understand classroom interaction;The social network related parameters of classroom interaction are calculated,and the interactive structure and classroom mode of the classroom are drawn according to the teaching mode in S-T analysis.This paper uses speaker-independent voiceprint recognition technology and social network analysis to quantify and characterize classroom interaction behavior,which can form an objective evaluation of classroom interaction and provide consistent discriminant indicators for classroom interaction to achieve horizontal and vertical contrasts.In addition,it is helpful to improve the credibility and scientificity of the data analysis process.And it is of great significance to embed the educational technology into the learning system to evaluate the learning process.