SelfVC: Voice Conversion With Iterative Refinement using Self Transformations

Computer Science

• Synthesis Model 100%
• Iterative Refinement 100%
• Information Loss 50%
• Self-Supervised Learning 50%
• Speaker Verification 50%
• Training Objective 50%
• Self-Supervised Representation Learning 50%

Keyphrases

• Speaker Similarity 33%
• Speech Representation 33%
• Linguistic Content 16%
• Zero-shot Voice Conversion 16%
• Self-Supervised Learning Representation 16%
• Speaker Intelligibility 16%