YuE2 covers frequently break: one mistimed word derails the whole song's lyrics
Gyramuur · reddit · 2026-09-17
A user reports open-source music model YuE2 works fine for text-to-audio in ComfyUI but fails often on covers: lyrics get wrong despite being explicitly written, or a single word sung early derails the timing of the rest of the song, sometimes degrading into gibberish or omitting entire bars.
They're unsure whether the bug lies in the model, ComfyUI, or sheetsage transcription, and are asking for workarounds.
More from Multimodal
- Viewers can't tell it's AI: Higgsfield video sparks authenticity drama — mallow610 · 2026-09-17
- AI-made Neymar version of STORM II — Stunning_Act_1530 · 2026-09-17
- Proposal: audio RefMods for MiniMax-H3, bringing IP-Adapter mechanics to voice timbre — ledadu · 2026-09-17
- QuiverAI logo generation impresses in hands-on test — jonathan_wilke · 2026-09-17
- Suno user slams pop stars' AI music claims, calls for abolishing copyright system — TinfoilTricorn · 2026-09-17
- One prompt drives Blender end-to-end: MiniMax Design + MCP workflow demo — iamfakhrealam · 2026-09-17