Personalization of CTC-Based End-to-End Speech Recognition Using Pronunciation-Driven Subword Tokenization.
Zhihong LeiErnest PusateriShiyi HanLeo LiuMingbin XuTim NgRuchir TravadiYouyuan ZhangMirko HannemannMan-Hung SiuZhen HuangPublished in: ICASSP (2024)
Keyphrases
- end to end
- speech recognition
- broadcast news
- automatic speech recognition
- spontaneous speech
- speech recognizer
- speech signal
- hidden markov models
- spoken document retrieval
- language model
- acoustic models
- congestion control
- high bandwidth
- admission control
- speaker independent
- wireless ad hoc networks
- speaker identification
- n gram
- named entities
- ad hoc networks
- internet protocol
- mobile devices
- multipath
- context aware
- spoken language
- application layer
- content delivery
- image coding
- multimedia