HiGNN-TTS: Hierarchical Prosody Modeling with Graph Neural Networks for Expressive Long-form TTS.
Dake GuoXinfa ZhuLiumeng XueTao LiYuanjun LvYuepeng JiangLei XiePublished in: CoRR (2023)
Keyphrases
- text to speech
- neural network
- speech synthesis
- prosodic features
- random walk
- directed graph
- back propagation
- word processing
- fuzzy logic
- multilayer perceptron
- hybrid models
- graphical representation
- graph theory
- neural network model
- pattern recognition
- genetic algorithm
- feed forward
- graph structure
- weighted graph
- multi layer
- fuzzy cognitive maps
- fault diagnosis
- graphical models
- image segmentation