A Wav2Vec2 model pre-trained only on Dutch encodes Dutch phonetic and lexical information better than English-only or larger multilingual models, and this improvement carries over to Dutch ASR.
Each SSL model is fine-tuned on Dutch read-aloud speech from the CGN (component o), using 78 hours of training data while reserving 10 hours each for de- velopment and testing
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
-
What do self-supervised speech models know about Dutch? Analyzing advantages of language-specific pre-training
A Wav2Vec2 model pre-trained only on Dutch encodes Dutch phonetic and lexical information better than English-only or larger multilingual models, and this improvement carries over to Dutch ASR.