Multi-accent deep neural network acoustic model with accent-specific top layer using the KLD-regularized model adaptation.
Yan HuangDong YuChaojun LiuYifan GongPublished in: INTERSPEECH (2014)
Keyphrases
- neural network
- spoken language
- speech recognition
- multi layer
- artificial neural networks
- domain specific
- neural network model
- least squares
- kullback leibler distance
- kullback leibler divergence
- back propagation
- high level
- self organizing maps
- fuzzy logic
- recurrent neural networks
- neural nets
- knn
- fuzzy neural network
- pattern recognition
- objective function