Domain Adaptation with External Off-Policy Acoustic Catalogs for Scalable Contextual End-to-End Automated Speech Recognition.
David M. ChanShalini GhoshAriya RastrowBjörn HoffmeisterPublished in: ICASSP (2023)
Keyphrases
- end to end
- speech recognition
- domain adaptation
- speech recognition systems
- speaker independent
- speech recognizers
- hidden markov models
- acoustic models
- multiple sources
- language model
- speech signal
- cross domain
- transfer learning
- speech recognizer
- speech synthesis
- semi supervised
- automatic speech recognition
- semi supervised learning
- pattern recognition
- labeled data
- speaker identification
- sentiment classification
- target domain
- unlabeled data
- acoustic features
- test data
- text mining
- unsupervised learning
- visual features
- training data
- feature extraction
- learning algorithm