End-to-End Spelling Correction Conditioned on Acoustic Feature for Code-Switching Speech Recognition.
Shuai ZhangJiangyan YiZhengkun TianYe BaiJianhua TaoXuefei LiuZhengqi WenPublished in: Interspeech (2021)
Keyphrases
- end to end
- speech recognition
- speech recognition systems
- spelling correction
- speaker independent
- discriminative training
- acoustic models
- speech recognizers
- hidden markov models
- automatic speech recognition
- context sensitive
- language model
- speech signal
- speech recognizer
- pattern recognition
- speaker identification
- english text
- speech synthesis
- word sense disambiguation
- noisy environments
- acoustic features
- feature vectors
- computer vision
- search queries
- information retrieval
- feature set
- classification accuracy
- feature extraction