Contrastive Self-Supervised Speaker Embedding With Sequential Disentanglement

Keyphrases

• Speaker Embedding 100%
• Disentanglement 100%
• Contrastive Learning 60%
• Learning Framework 40%
• Content Factors 40%
• Speaker Identity 40%
• Linguistic Content 40%
• Speaker Factor 40%
• Sequential Variational Autoencoder 40%
• Speech Signal 20%
• Embedding Space 20%
• Speech Segments 20%
• Language Mismatch 20%
• Supervised Contrastive Learning 20%
• VoxCeleb 20%
• Contrastive Loss 20%
• Discriminative Embedding 20%

Computer Science

• Contrastive Learning 100%
• Learning Framework 66%
• Variational Autoencoder 66%
• Embedding Space 33%
• Self-Supervised Learning 33%
• Invariant 33%
• Contrastive Loss 33%