Fine-tuning MMS-1B with auxiliary LID CTC loss on augmented data gives 86.9% LID accuracy and 15.6% CER on the ML-SUPERB 2.0 development set.
2 Models are evaluated on the development sets due to the unavailability of the test set
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SD 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Improving Multilingual Speech Models on ML-SUPERB 2.0: Fine-tuning with Data Augmentation and LID-Aware CTC
Fine-tuning MMS-1B with auxiliary LID CTC loss on augmented data gives 86.9% LID accuracy and 15.6% CER on the ML-SUPERB 2.0 development set.