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: INTERSPEECH (2023)
Keyphrases
- end to end
- automatic speech recognition
- word error rate
- word recognition
- speech recognition
- conversational speech
- recognition errors
- compound words
- speech signal
- congestion control
- word level
- hidden markov models
- speech retrieval
- spontaneous speech
- admission control
- broadcast news
- speech recognizer
- noisy environments
- handwriting recognition
- feature selection
- error rate
- language model
- feature set
- computer vision
- application layer
- word segmentation
- multimedia
- speech sounds