AviadDahan/SepGen-Separation
Audio-to-Audio • Updated • 208 • 1
SepGen
SepGen: Multi-Stem Audio-Video Separation and Generation in a Single Model. Separation (sep-12k) and generation (gen-3k) checkpoints.
Note sep-12k: separation
AviadDahan/SepGen-Generation
Text-to-Video • Updated • 69 • 1Note gen-3k: generation and separation
AviadDahan/SepGen-Dataset
Viewer • Updated • 4.85k • 131Note training manifests (captions, boxes, provenance)
-
SepGen: Multi-Stem Audio-Video Separation and Generation in a Single Model
Paper • 2610.11361 • Published • 47 -
SepGen Walk-through
🎧1Walk into a scene and hear each separated sound in 3D