Improving CTC-based speech recognition via knowledge transferring from pre-trained language models.
Keqi DengSongjun CaoYike ZhangLong MaGaofeng ChengJi XuPengyuan ZhangPublished in: CoRR (2022)
Keyphrases
- speech recognition
- language model
- language modeling
- pre trained
- n gram
- information retrieval
- document retrieval
- speech signal
- probabilistic model
- automatic speech recognition
- query expansion
- pattern recognition
- test collection
- word error rate
- mixture model
- retrieval model
- handwriting recognition
- speech recognition systems
- hidden markov models
- cross lingual
- generative model
- multi modal
- active learning
- feature extraction
- image processing