VAENAR-TTS: Variational Auto-Encoder Based Non-AutoRegressive Text-to-Speech Synthesis.
Hui LuZhiyong WuXixin WuXu LiShiyin KangXunying LiuHelen MengPublished in: Interspeech (2021)
Keyphrases
- autoregressive
- text to speech
- text to speech synthesis
- non stationary
- moving average
- bit rate
- gaussian markov random field
- image segmentation
- random fields
- word processing
- autoregressive model
- motion estimation
- sar images
- optical flow
- spectrum analysis
- random field models
- image processing
- machine learning
- maximum likelihood
- bayesian networks
- model selection
- probabilistic model
- multiresolution
- multiscale