Please use this identifier to cite or link to this item: https://repository.isls.org/handle/1/10538
Title: Speech-based Automatic Classroom Activity Detection
Authors: Shahrokhian, Bahar
Zhou, David
Vanlehn, Kurt
Keywords: CSCL
Issue Date: 2024
Publisher: International Society of the Learning Sciences
Citation: Shahrokhian, B., Zhou, D., & Vanlehn, K. (2024). Speech-based Automatic Classroom Activity Detection. In Clarke-Midura, J., Kollar, I., Gu, X., & D'Angelo, C. (Eds.), Proceedings of the 17th International Conference on Computer-Supported Collaborative Learning - CSCL 2024 (pp. 301-304). International Society of the Learning Sciences.
Abstract: Developing an automatic inexpensive, non-intrusive approach that can provide timely feedback to teachers in a noisy co-located classroom remains a difficult task. The first step is to detect class activities robustly in this noisy setting and then build a more fine-grained classifier on top of that. In this work, we trained a machine learning model that takes in speech data captured from low-cost microphones and recorders, while teachers were conducting the class using an orchestration system. We used acoustic, pre-trained representation(acoustic), and contextual features as input to a supervised machine learning algorithm to classify classroom activity into 3 categories (whole-class, group work, and others). The combination of all the features yielded the best result (F1-measure=0.91) on previously unseen data.
Description: Short Paper
URI: https://doi.org/10.22318/cscl2024.886155
https://repository.isls.org//handle/1/10538
Appears in Collections:ISLS Annual Meeting 2024

Files in This Item:
File SizeFormat 
CSCL2024_301-304.pdf338.99 kBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.