DNN speaker adaptation using parameterised sigmoid and ReLU hidden activation functions.
Chao ZhangPhilip C. WoodlandPublished in: ICASSP (2016)
Keyphrases
- activation function
- speaker adaptation
- neural network
- speech recognition
- artificial neural networks
- feed forward
- hidden layer
- back propagation
- neural nets
- learning rate
- maximum likelihood
- automatic speech recognition
- basis functions
- speaker independent
- network architecture
- multilayer perceptron
- training phase
- sigmoid function
- radial basis function
- hidden markov models
- input output
- artificial intelligence
- rbf neural network
- fuzzy neural network
- multi layer perceptron
- language model
- speech recognizer
- genetic algorithm