Bayesian Unsupervised Batch and Online Speaker Adaptation of Activation Function Parameters in Deep Models for Automatic Speech Recognition.
Zhen HuangSabato Marco SiniscalchiChin-Hui LeePublished in: IEEE ACM Trans. Audio Speech Lang. Process. (2017)
Keyphrases
- automatic speech recognition
- speaker adaptation
- speech recognition
- maximum likelihood
- activation function
- autoregressive
- speech signal
- sigmoid function
- hidden markov models
- probabilistic model
- back propagation
- random fields
- learning rate
- neural network
- artificial neural networks
- hidden layer
- non stationary
- fuzzy logic
- noisy environments
- speech recognizer
- bayesian networks
- computer vision