Improving Deep Neural Network Based Speech Synthesis through Contextual Feature Parametrization and Multi-Task Learning.
Zhengqi WenKehuang LiZhen HuangChin-Hui LeeJianhua TaoPublished in: J. Signal Process. Syst. (2018)
Keyphrases
- multi task learning
- speech synthesis
- multi task
- speech recognition
- multiple tasks
- text to speech
- multi label image annotation
- learning tasks
- multitask learning
- gaussian processes
- neural network
- learning problems
- theoretical analysis
- high order
- transfer learning
- learning algorithm
- learning models
- image features
- hidden markov models
- machine learning algorithms
- context sensitive
- inductive learning
- pattern recognition
- feature set
- feature vectors
- reinforcement learning
- image processing
- feature selection