A Method For Estimating The Grouping Of Participants In Classroom Group Work Using Only Audio Information

This paper proposes a novel method for estimating which microphone belongs to the same group in a situation where there are multiple discussion groups in one room, using only audio information. The assumption is that each member wears one close-talk microphone, and that the audio is recorded on their own audio track. Each microphone records the main speech of the associated speaker, as well as the speech of neighboring others that have leaked into the microphone. If the neighboring speaker's leaked speech is coming in clearly, the neighboring speaker can be in proximity. An undirected network is constructed with speakers as nodes and the degrees of leaked speech as the edge weights. Given a target number of discussion groups in a room, network clustering can be applied to obtain subgroup information about which audio tracks belong to the same group. An evaluation experiment was conducted using audio data recorded in workgroup classes at the actual junior high school. In this experiment, the average Rand index of the grouping was 0.995, confirming that practical accuracy can be obtained.