Deep Feature CycleGANs: Speaker Identity Preserving Non-Parallel Microphone-Telephone Domain Adaptation for Speaker Verification.
Saurabh KatariaJesús VillalbaPiotr ZelaskoLaureano Moro-VelázquezNajim DehakPublished in: Interspeech (2021)
Keyphrases
- speaker verification
- domain adaptation
- speaker diarization
- noisy environments
- speaker recognition
- prosodic features
- audio visual
- cross domain
- multiple sources
- semi supervised
- sentiment classification
- multilayer perceptron
- feature representation
- labeled data
- semi supervised learning
- feature vectors
- unlabeled data
- image features
- test data
- machine learning
- emotion recognition
- automatic speech recognition
- information retrieval
- transfer learning
- natural language processing
- target domain
- multi modal
- co training
- pattern recognition
- training data
- visual information