Preprint
Automatically Detecting Confusion and Conflict During Collaborative Learning Using Linguistic, Prosodic, and Facial Cues
Yingbo Ma et al., arXiv, 2024
Read the originalRead on October 17, 2026
- What it is
- A study of elementary school learners' programming dialogues in classrooms. The authors ran the audio through three speech recognition engines and compared the error rates before using the Google transcripts for their model.
- How much weight it carries
- A preprint, not yet peer-reviewed when checked. The recordings are children talking in classrooms, a narrow case, and the paper reports its error rates on its own scale.
Facts taken from this source
- In a 2024 study of children's classroom dialogue, Google Speech-to-Text had the lowest mean error rate of the three engines compared and the lowest variation, which the authors describe as the most robust to noise.
“Google Speech-to-Text achieved the most robust performance in the face of substantial audio noise, with the lowest mean WER score of 0.78” (Table 2)