Improving Frame-level Classifier for Word Timings with Non-peaky CTC in End-to-End Automatic Speech Recognition.
Xianzhao ChenYist Y. LinKang WangYi HeZejun MaPublished in: CoRR (2023)
Keyphrases
- end to end
- automatic speech recognition
- word recognition
- word error rate
- speech recognition
- conversational speech
- recognition errors
- compound words
- hidden markov models
- speech signal
- spontaneous speech
- admission control
- speech retrieval
- congestion control
- broadcast news
- word level
- speech recognizer
- noisy environments
- feature selection
- application layer
- n gram
- error rate
- pattern recognition
- transport layer