Subspace Based Sequence Discriminative Training of LSTM Acoustic Models with Feed-Forward Layers.
Lahiru SamarakoonBrian MakAlbert Y. S. LamPublished in: ISCSLP (2018)
Keyphrases
- feed forward
- discriminative training
- acoustic models
- recurrent neural networks
- hidden markov models
- back propagation
- maximum likelihood
- probabilistic interpretation
- speech recognition
- artificial neural networks
- posterior probability
- neural network
- automatic speech recognition
- support vector machine
- co training
- speech recognizer
- dimensionality reduction
- low dimensional
- broadcast news
- pattern recognition
- feature space
- high dimensional
- language model
- principal component analysis
- speaker independent
- fuzzy logic