An encoder-based CTC model that predicts whole contextual phrases as single bias tokens, with confidence-based replacement, cuts contextual-phrase WER by about 72% on LibriSpeech and 76% on WenetSpeech.
Experimental setup We train CTC/AED models as the baseline and the pre-trained ASR model, using the Wenet toolkit [25]
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
-
Contextualized Automatic Speech Recognition with Dynamic Vocabulary Prediction and Activation
An encoder-based CTC model that predicts whole contextual phrases as single bias tokens, with confidence-based replacement, cuts contextual-phrase WER by about 72% on LibriSpeech and 76% on WenetSpeech.