Improving speech recognition using limited accent diverse British English training data with deep neural networks.
Maryam NajafianSaeid SafaviJohn H. L. HansenMartin J. RussellPublished in: MLSP (2016)
Keyphrases
- spoken language
- neural network
- training data
- speech recognition
- training process
- broadcast news
- speech recognizer
- automatic speech recognition
- text to speech
- spontaneous speech
- english text
- pattern recognition
- speech recognition technology
- dialogue system
- acoustic models
- decision trees
- language processing
- speech signal
- learning algorithm
- classification accuracy
- test data
- supervised learning
- natural language
- machine translation
- prior knowledge
- english language
- artificial neural networks
- recurrent neural networks
- feed forward
- test set
- data sets
- training set
- fuzzy logic
- speech synthesis
- genetic algorithm
- semantic analysis
- neural nets
- classification models
- language learning
- training samples
- query translation
- hidden markov models
- answer questions
- multi layer
- word processing
- back propagation
- training examples
- labeled data
- language model
- support vector machine