Exploring Branchformer-Based End-to-End Speaker Diarization with Speaker-Wise VAD Loss

INIS

• losses 100%
• datasets 100%
• performance 50%
• comparative evaluations 50%
• margins 50%
• simulation 50%
• errors 50%
• speech 50%
• streams 50%
• attractors 50%
• benchmarks 50%
• transformers 50%

Computer Science

• Speaker Diarization 100%
• Traditional Method 50%
• Experimental Result 50%
• Self-Attention Mechanism 50%