Hypothesis Correction Based on Semi-character Recurrent Neural Network for End-to-end Speech Recognition.
Yuuki TachiokaPublished in: APSIPA (2019)
Keyphrases
- speech recognition
- end to end
- recurrent neural networks
- text localization and recognition
- neural network
- feed forward
- hidden markov models
- language model
- speech signal
- pattern recognition
- artificial neural networks
- automatic speech recognition
- congestion control
- speech processing
- speech recognition technology
- speech recognition systems
- speech recognizer
- noisy environments
- recurrent networks
- speech synthesis
- speaker dependent
- speaker independent
- handwriting recognition
- information retrieval
- speaker identification
- optical character recognition
- maximum likelihood
- artificial intelligence
- chinese characters
- probabilistic model
- document images
- motion estimation