Listen Attentively, and Spell Once: Whole Sentence Generation via a Non-Autoregressive Architecture for Low-Latency Speech Recognition.
Ye BaiJiangyan YiJianhua TaoZhengkun TianZhengqi WenShuai ZhangPublished in: CoRR (2020)
Keyphrases
- speech recognition
- autoregressive
- low latency
- non stationary
- hidden markov models
- real time
- language model
- high speed
- automatic speech recognition
- speech recognition systems
- speech signal
- pattern recognition
- high throughput
- random fields
- highly efficient
- multiscale
- sar images
- virtual machine
- stream processing
- natural language
- data mining
- multiresolution
- linear prediction
- machine learning
- information retrieval
- database systems
- markov random field