USAD 2.0: Scaling Representation Distillation for Universal Audio Understanding
Paper • 2606.06444 • Published • 3
USAD2
USAD 2.0: Scaling Representation Distillation for Universal Audio Understanding
-
-
MIT-SLS/USAD2-Small
Feature Extraction • 24.8M • Updated • 20 -
MIT-SLS/USAD2-Base
Feature Extraction • 97.2M • Updated • 49 -
MIT-SLS/USAD2-Large
Feature Extraction • 0.3B • Updated • 185 • 1 -
MIT-SLS/USAD2-XLarge
Feature Extraction • 0.7B • Updated • 18 • 1 -
MIT-SLS/USAD2-Large-Plus
Feature Extraction • 0.3B • Updated • 143 • 1 -
MIT-SLS/USAD2-XLarge-Plus
Feature Extraction • 0.7B • Updated • 35 • 1 -
MIT-SLS/USAD2-XXLarge-Plus
Feature Extraction • 1B • Updated • 45 • 1