Domain Adaptation Based on Mixture of Latent Words Language Models for Automatic Speech Recognition.
Ryo MasumuraTaichi AsamiTakanobu ObaHirokazu MasatakiSumitaka SakauchiAkinori ItoPublished in: IEICE Trans. Inf. Syst. (2018)
Keyphrases
- target domain
- domain adaptation
- language model
- automatic speech recognition
- word error rate
- speech recognition
- n gram
- transfer learning
- cross domain
- word recognition
- training data
- language modeling
- labeled data
- out of vocabulary
- probabilistic model
- test collection
- speech signal
- information retrieval
- document retrieval
- broadcast news
- unlabeled data
- query expansion
- document level
- hidden markov models
- expectation maximization
- semi supervised
- active learning
- vector space model
- retrieval model
- semi supervised learning
- pseudo relevance feedback
- word level
- error rate
- sentiment classification
- relevance model