New: ComfyUI-MiniMax-Music-Production-Toolkit, for prompt generation and more
Hey, for anyone who has trouble creating the rather complex prompts required by MiniMax Music 3: I’ve created an open-source ComfyUI MiniMax Music Production Toolkit that automates a large part of the workflow. This consists of a custom node package for ComfyUI, and includes a well designed workflow.
It includes LLM-based prompt generation for Caption, Lyrics, Title and Cover Image prompts, reusable prompt libraries, metadata/tagging, automatic album artwork generation, FLAC/MP3 output, audio restoration/upscaling (minimax has some weaknesses), loudness preparation and batch-friendly file management.
The important part for batch generation is that every run can create a genuinely different result: Composition, arrangement, title, lyrics and cover artwork, all from a simple genre/style prompt.
I tested it on RTX 5060 Ti 16 GB VRAM + 64 GB DDR4 RAM, with a quite old Intel 9400, including local prompt generation with Qwen3.8-27B. It works very well on this setup. A ~4-5 minute track takes roughly 8-12 minutes to generate in my workflow.
GitHub:
https://github.com/jplenio/ComfyUI-MiniMax-Music-Production-Toolkit
Maybe this helps some of you get more consistent results from MiniMax Music 3 without having to manually construct all the structured prompts every time.
