Many-to-Many Voice Conversion based on Bottleneck Features with Variational Autoencoder for Non-parallel Training Data.
Yanping LiKong-Aik LeeYougen YuanHaizhou LiZhen YangPublished in: APSIPA (2018)
Keyphrases
- training data
- classification accuracy
- classification models
- prior knowledge
- low level
- decision trees
- training examples
- test data
- false positives
- extracted features
- data sets
- feature vectors
- shared memory
- feature values
- feature extraction
- feature weights
- variational methods
- supervised learning
- spatial information
- feature set
- pairwise
- training set
- image segmentation
- learning algorithm