4 days ago
-
Release Audio2Sph-L 543k weights and document four-second SourceBank generation
-
Refine reader documentation and output contracts
Red-team revision for reader order, required prerequisites, output schemas, metric definitions, training configuration, and terminology; refresh demo audio to 48 kHz. -
Finalize evaluation versions and six-metric results
-
Release Audio2Sph and Panoramic Decoder checkpoint
Add the paper class-agnostic Audio2Sph + Panoramic Decoder checkpoint, checkpoint card, native inference documentation, and TAU/Sony comparison demos.
-
Add SAID documentation and media
