A two-stage cascade of self-supervised training followed by Meta Pseudo-Labeling improves unsupervised domain adaptation for Greek ASR.
Pre-trained model: For our base model, we utilize XLSR-53 [23] , a state-of-the-art pre-trained speech model developed on the Wav2Vec 2.0 [20] architecture
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
MSDA: Combining Pseudo-labeling and Self-Supervision for Unsupervised Domain Adaptation in ASR
A two-stage cascade of self-supervised training followed by Meta Pseudo-Labeling improves unsupervised domain adaptation for Greek ASR.