Iterative Music Separation: General Audio Separation Framework Open-Sourced with Training Code
affige_yang · x · 2026-09-09
Researcher yukaraikemiya announced a general framework for iterative audio separation, focused on music source separation.
- Full training code and pretrained weights are open-sourced
- Significant improvements using SOTA architectures
- Project page and GitHub repo are available for reproduction
More from Multimodal
- SuperSplat browser viewer showcases stunning user-uploaded 3D Gaussian captures — willeastcott · 2026-09-09
- 5 ChatGPT Image Prompts That Turn Casual Photos Into Pro Shots (Face Preserved) — TawohAwa · 2026-09-09
- Sculpt Canvas Brings 10-Brush 3D Sculpting Inside ComfyUI Graphs — skbphy · 2026-09-09
- 'Generate for the Edit' Accepted to SIGGRAPH Asia 2026 Posters — ring_hyacinth · 2026-09-09
- 10 Hours, 21 Revisions, 1.3M Polygons: Artist Builds Vintage Gym in Blender via GPT-6 Astra — petewoodbridge · 2026-09-09
- Recreating Ghostbusters scene locally with MiniMax H3 on a 16GB RTX 4070 Ti — darthfurbyyoutube · 2026-09-09