Building Bilingual and Code-Switched Voice Conversion with Limited Training Data Using Embedding Consistency Loss.
Yaogen YangHaozhe ZhangXiaoyi QinShanshan LiangHuahua CuiMingyang XuMing LiPublished in: CoRR (2021)
Keyphrases
- training data
- learning algorithm
- prior knowledge
- test data
- data sets
- training examples
- source code
- information retrieval
- training process
- machine translation
- class labels
- semi supervised learning
- classification accuracy
- domain knowledge
- supervised learning
- labeled data
- support vector machine
- unlabeled data
- high dimensional
- training set
- vector space
- noisy data
- generalization error
- cross language
- support vector
- query translation
- statistical machine translation
- word alignment