| 中文题名： | 基于音频特征的课堂出声教学活动自动分类 |
| 姓名： | |
| 保密级别： | 公开 |
| 论文语种： | 中文 |
| 学科代码： | 080901 |
| 学科专业： | |
| 学生类型： | 学士 |
| 学位： | 理学学士 |
| 学位年度： | 2022 |
| 学校： | 北京师范大学 |
| 校区： | |
| 学院： | |
| 第一导师姓名： | |
| 第一导师单位： | |
| 提交日期： | 2022-05-24 |
| 答辩日期： | 2022-05-13 |
| 外文题名： | Automatic classification of classroom out-of-sound teaching activities based on audio features |
| 中文关键词： | |
| 外文关键词： | |
| 中文摘要： |
近年来，基于人工智能的课堂教学自动评价随着“智慧教育”的出现而成为研究热点，利用信息技术实现教学评价自动化已成为当前研究的主流趋势。目前使用音频特征仅能够自动区分教师行为和学生行为，实现课堂教学过程的S-T自动编码与分析，无法对更多的教学活动进行识别。本文通过提取音频信号的显著特征训练卷积神经网络，实现了教师讲授、影视视听、学生单人发言、小组讨论、师生言语互动、齐读、与课堂教学内容无关的其它声音等七类课堂出声活动的自动分类，可将一节完整的教学音频以连续3s的间隔依次自动标注出活动类别，为实现教学评价自动化提供全新的思路和相应的技术支持，助力“智慧教育”。
﹀
|
| 外文摘要： |
In recent years, automatic evaluation of classroom teaching based on artificial intelligence has become a research hotspot with the emergence of "smart education", and the use of information technology to automate teaching evaluation has become the mainstream trend of current research. At present, the use of audio features can only automatically distinguish teacher behavior from student behavior and achieve automatic S-T coding and analysis of classroom teaching process, but cannot identify more teaching activities. In this paper, we train convolutional neural networks by extracting significant features of audio signals to achieve automatic classification of seven categories of classroom audio activities, such as teacher lecture, video and audio, single student speech, group discussion, teacher-student verbal interaction, unison reading, and other sounds unrelated to classroom teaching content, which can automatically label the activity categories of a complete teaching audio session in a continuous 3s interval, and provide a new idea and corresponding technical support for automating teaching evaluation. It provides a new idea and corresponding technical support to help "smart education".
﹀
|
| 参考文献总数： | 21 |
| 插图总数： | 12 |
| 插表总数： | 10 |
| 馆藏号： | 本080901/22033 |
| 开放日期： | 2023-05-24 |
谷歌
