HMM-Free Encoder Pre-Training for Streaming RNN Transducer.
Lu HuangJingyu SunYufeng TangJunfeng HouJinkun ChenJun ZhangZejun MaPublished in: Interspeech (2021)
Keyphrases
- hidden markov models
- recurrent neural networks
- nearest neighbor
- streaming data
- bit rate
- test set
- low complexity
- training set
- speech recognition
- artificial neural networks
- motion estimation
- real time
- training samples
- video compression
- concept drift
- power consumption
- training examples
- supervised learning
- data streams
- feature selection