Kaizen: Continuously improving teacher using Exponential Moving Average for semi-supervised speech recognition.
Vimal ManoharTatiana LikhomanenkoQiantong XuWei-Ning HsuRonan CollobertYatharth SarafGeoffrey ZweigAbdelrahman MohamedPublished in: CoRR (2021)
Keyphrases
- speech recognition
- moving average
- semi supervised
- autoregressive
- hidden markov models
- language model
- speech processing
- speech synthesis
- speech signal
- speech recognizer
- automatic speech recognition
- speech recognition technology
- pattern recognition
- co training
- speaker dependent
- speech recognition systems
- audio visual speech recognition
- learning process
- non stationary
- noisy environments
- speech recognizers
- active learning
- speaker identification
- isolated word