Adding more training data improves voice-type classification more than architecture or representation changes, which plateau near 50% average F-score, still below human annotator agreement of about 70%.
Despite these challenges, this technology holds transformative potential for educational interventions and insights into children’s lan- guage learning mechanisms
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
eess.AS 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Challenges in Automated Processing of Speech from Child Wearables: The Case of Voice Type Classifier
Adding more training data improves voice-type classification more than architecture or representation changes, which plateau near 50% average F-score, still below human annotator agreement of about 70%.