Sa-Paraformer: Non-Autoregressive End-To-End Speaker-Attributed ASR.
Yangze LiFan YuYuhao LiangPengcheng GuoMohan ShiZhihao DuShiliang ZhangLei XiePublished in: ASRU (2023)
Keyphrases
- end to end
- autoregressive
- automatic speech recognition
- speech recognition
- simulated annealing
- non stationary
- moving average
- speech signal
- hidden markov models
- gaussian markov random field
- random fields
- admission control
- noisy environments
- random field models
- sar images
- application layer
- scalable video
- high resolution
- internet protocol
- spontaneous speech
- multiscale