Simpleflat: A Simple Whole-Network Pre-Training Approach for RNN Transducer-Based End-to-End Speech Recognition.
Takafumi MoriyaTakanori AshiharaTomohiro TanakaTsubasa OchiaiHiroshi SatoAtsushi AndoYusuke IjimaRyo MasumuraYusuke ShinoharaPublished in: ICASSP (2021)
Keyphrases
- end to end
- speech recognition
- wireless ad hoc networks
- congestion control
- internet protocol
- wall street journal corpus
- transport layer
- hidden markov models
- isolated word
- packet loss rate
- pattern recognition
- speech processing
- language model
- automatic speech recognition
- speech signal
- noisy environments
- speech recognizer
- speaker identification
- speech recognition technology
- speech synthesis
- information retrieval
- speech recognition systems
- acoustic models
- ad hoc networks
- admission control
- speaker independent
- application layer
- network resources
- recurrent neural networks
- speaker dependent
- neural network
- training set
- multimedia