WanSong v1.0 Generates 5-Minute Songs via Pure Diffusion
July 15, 2026
WanSong is a pure diffusion-based model for generating high-fidelity, multilingual songs up to five minutes in duration. It outputs dual stems—vocals and background music—in a single pass and utilizes step-distillation for faster inference.
HOW THIS AFFECTS YOU
●
builderYou can integrate high-fidelity, long-form audio generation into production pipelines via distilled diffusion.
●
designerThis allows for more granular control over song structure through separate vocal and instrumental stems.