Research
AMBIT framework generates and edits First-Order Ambisonics spatial audio
AMBIT couples a Transfusion backbone with executable scene plans and self-distillation to generate and edit spatial audio. The paper reports gains over compared baselines in content, spatial quality and source preservation across speech, music and sound.
Every outlet that ran it 1 article
Hugging Face PapersResearchLead
Spatial Audio Generation and Editing with Executable Plans and Self-Distilling Transfusion