Minimum word error training of long short-term memory recurrent neural network language models for speech recognition.
Takaaki HoriChiori HoriShinji WatanabeJohn R. HersheyPublished in: ICASSP (2016)
Keyphrases
- speech recognition
- recurrent neural networks
- language model
- wall street journal corpus
- long short term memory
- n gram
- language modeling
- recurrent networks
- speech recognizer
- feed forward
- neural network
- speech recognition systems
- word error rate
- artificial neural networks
- translation model
- probabilistic model
- query expansion
- information retrieval
- speech recognizers
- mixture model
- speech synthesis
- document retrieval
- error rate
- retrieval model
- speech signal
- automatic speech recognition
- test collection
- computer vision
- noisy environments
- relevance model
- hidden markov models
- handwriting recognition
- query terms
- relevance feedback
- text retrieval