Unsupervised Style and Content Separation by Minimizing Mutual Information for Speech Synthesis.
Ting-Yao HuAshish ShrivastavaOncel TuzelChandra DhirPublished in: CoRR (2020)
Keyphrases
- multimedia
- speech synthesis
- mutual information
- speech recognition
- information theoretic
- image registration
- text to speech
- feature selection
- vocal tract
- similarity measure
- machine learning
- medical image registration
- information gain
- prosodic features
- conditional mutual information
- metadata
- digital libraries
- unsupervised learning
- multi modal image registration
- information theoretic measures
- mutual information maximization
- competitive learning
- data driven
- language model
- semi supervised
- social networks