ComfyUI-Olm-YuE2: Staged YuE2 Music Generation With an Editable Score Workflow
imlo2 · reddit · 2026-09-11
A developer released ComfyUI-Olm-YuE2, integrating the newly released open music model YuE2 into ComfyUI with far more than a single Generate button:
- Staged pipeline: generation split into Plan > Semantic > Synthesize > Decode, with intermediate results inspectable, savable, reusable and branchable.
- Optional score UI: sidebar rendered notation, raw ABC score view, an ABC editor node, and the ability to generate a plan, edit the score, then continue generation from the edited version.
- Input/output: style + lyrics in, 48 kHz stereo songs out.
- Engineering details: reuses ComfyUI's Torch/Transformers, needing only tiktoken and accelerate on top; includes measured VRAM numbers (tested on RTX 5090) and an experimental offload; weights not bundled. Environment: CUDA 13.0, Python 3.13.
Open source on GitHub; the author is seeking feedback from 16/24 GB GPU users.
Related event: YuE2 Music Generation Comes to ComfyUI with Editable Scores(2 posts)→
More from Multimodal
- I tested 11 video upscaling methods for AI video: local, Topaz and cloud compared — lopas99 · 2026-10-28
- AI Horde adds Moss-Image for free, no-account image generation on volunteer GPUs — Wanabeelee2 · 2026-10-28
- AI band Techhalla reimagines Slipknot's 'Wait and Bleed' as a 1971 soul performance — Mother-Pear-2392 · 2026-10-28
- Inside the Hailuo H3 Max retro-MV workflow: 15-second chunks, one reference image, multi-shot prompts — techhalla · 2026-10-28
- Open-source pipeline turns Seedance AI video into real PBR textures on your 3D mesh — OdinLovis · 2026-10-28
- AI-generated cyberpunk short 'VHS Sirens' shows off consistent characters and neon city detail — brythehuman · 2026-10-28