Adaptation of Hybrid ANN/HMM Models Using Linear Hidden Transformations and Conservative Training.
Roberto GemelloFranco ManaStefano ScanzioPietro LafaceRenato De MoriPublished in: ICASSP (1) (2006)
Keyphrases
- hidden markov models
- artificial neural networks
- linear model
- neural network
- neural network model
- baum welch
- ann models
- nonlinear models
- structured prediction
- markov models
- linear models
- closed form
- parameter estimation
- training samples
- model selection
- prior knowledge
- feed forward
- speech recognition
- feed forward neural networks
- machine learning