A MAML classifier on L2-normalized Whisper audio features achieves 78.98 to 85.22 percent accuracy for cross-lingual abuse detection in ten Indian languages using only 50 to 200 labeled clips per language.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Towards Cross-Lingual Audio Abuse Detection in Low-Resource Settings with Few-Shot Learning
A MAML classifier on L2-normalized Whisper audio features achieves 78.98 to 85.22 percent accuracy for cross-lingual abuse detection in ten Indian languages using only 50 to 200 labeled clips per language.