S-T classifying method based on voiceprint recognition, S-T classifying device based on voiceprint recognition and equipment terminal

Abstract

The invention discloses an S-T classifying method based on voiceprint recognition, an S-T classifying device based on voiceprint recognition and an equipment terminal. The method comprises the following steps of building a target voiceprint standard model according to voiceprint features extracted from target voice sample data; dividing classroom voice data to be detected into different voice fragments through discourse detection, wherein each voice fragment includes voice data of a single speaker; matching each voice fragment with the target voiceprint standard model; calculating similarity scores; and judging the voice fragments with the scores higher than a preset score threshold value to be target voice. The S-T classifying method has the advantages that sequential sampling is performed in time dimension on the basis of behaviors; through the discourse detection, the voice fragments of respective discourse behaviors of teachers and students are divided; the teaching mode is analyzed according to collected data; behavior data in the teaching process is shown in a multi-dimensional and visual way; the accuracy during data collection is higher; the possibility of partial behaviordata loss during sampling is greatly reduced; and the classification accuracy in the continuous model matching process is improved.

Description

technical field [0001] The present invention belongs to the technical field of educational information, and more specifically relates to an S-T classification method, device and equipment terminal based on voiceprint recognition. Background technique [0002] With the continuous deepening of curriculum reform, teaching evaluation has become an important part of examining classroom teaching effectiveness. Discourse, as the main medium of knowledge dissemination in the classroom, runs through the entire teaching activity. The process evaluation of teaching activities based on classroom discourse can provide insight into teaching models and strategies, provide scientific data support for teaching reflection, and help teachers improve their professionalism. [0003] Teaching activities are a complex process. The evaluation of teaching cannot be based on the subjective feelings of researchers alone. It needs to be based on scientific scales combined with real-time observations o...

AI Technical Summary

Problems solved by technology

[0007] Aiming at at least one defect or improvement requirement of the prior art, the present invention provides an S-T classification method, device and equipment terminal based on voiceprint recognition. Firstly, the speaking time segments of teachers and students are separated according to the classroom voice, so as to establish the speaker identity, and then match all the separated speech segments with the pre-established teacher voiceprint standard model one by one and calculate the similarity score to judge whether each speech segment belongs to a teacher or a student; this method will be based on behavior sampling and computer intelligent recognition Fusion to improve the sampling accuracy of computer intelligent recognition and the accuracy of teaching analysis results; the analysis process is completely intelligent without manual participation, so it can be popularized and used in conventional classrooms; its purpose is to solve the limitations of application scenarios in existing technologies , the extracted behavior data is inaccurate or missing, and the accuracy of teacher and student behavior classification is poor

Images