AI Music Generator
Creates background music, jingles or full songs with lyrics from a description
by ElevenLabs
MiniMax Music 3 Text to Music
by Comfy OrgWorkflowMake video or audio
Makes a full song with vocals from your lyrics and a style description on your computer
About MiniMax Music 3 Text to Music
An official ComfyUI workflow that makes a full song with vocals from your lyrics and a style description, using MiniMax's Music 3 model on your computer. It saves the song as an audio file, up to about 5 minutes.
For Music producers sketching demosCreators who need original music
Made by Comfy Org
Comfy Org, the San Francisco company behind the open-source ComfyUI app, publishes this template. The music model itself comes from MiniMax.
- Stars
- ★ 1.4k
- License
- MIT
- Cost
- Free
- Level
- Intermediate
- Platform
- comfyui
Overview
This is an official ComfyUI workflow template for MiniMax Music 3, an open music model from MiniMax. You enter lyrics with section tags such as [verse] and [chorus], plus a description of the style, mood, vocals and arrangement, and it generates a complete song and saves it as an audio file. Songs default to 2 minutes and can run to about 5 minutes. It suits creators who want original demo songs or background music without an online music service.
It makes complete songs with sung vocals and a verse-chorus structure from open model files, so there is no per-song fee.
Try these first
Start with the first one: it is a small task you can check easily.
- Make a 60-second song from these four lines with a fixed seed: [paste lyrics]. Keep [verse] and [chorus] tags.First try · about 14 GB download
- Global Metadata: lo-fi hip-hop, 78 BPM, late-night mood. Vocal Details: soft, breathy vocal. Arrangement: warm piano, vinyl crackle.
- Keep the same caption and lyrics, change only the seed and make a second take.
- Set the length to 180 seconds and add [intro], [bridge] and [outro] sections.needs more memory
Example prompts written by Agent Flows Hub.
How it works
From the creator's documentation.
| Step | What happens | Data goes to |
|---|---|---|
| 1 · Model files | ComfyUI lists the model, text encoder and audio decoder files to download, about 14 GB | Your computer |
| 2 · Caption | You describe the style, mood, vocals and arrangement | Your computer |
| 3 · Lyrics | Your lyrics, with tags such as [verse] and [chorus] that set the song's structure | Your computer |
| 4 · Generate | MiniMax Music 3 makes the song on your graphics card; 2 minutes by default, up to about 5 | Your computer |
| 5 · Save | ComfyUI saves the song as an audio file; commercial use has license conditions | Your computer |
Good to know
Turn your own lyrics and a written style description into a finished song. The template loads the MiniMax Music 3 model files, encodes the caption and lyrics, generates the audio and saves it.
Limits
- The model files use the MiniMax-Music3 Community License: a commercial product or service that uses them must show the name MiniMax-Music3, and very large companies need written permission from MiniMax.
- Longer songs need more time and graphics memory. The template offers a smaller int8 model file and tiled decoding for cards with less memory.
Notes by Agent Flows Hub (original project name: audio_minimax_music_3 (ComfyUI template)), written from the creator's documentation, not from running it ourselves. MiniMax Music 3: pinned ComfyUI template · MiniMax-Music3 Community License
Stuck setting it up?
Tell us what happened. We take a look first; if it needs paid work, you get an estimate before anything starts.
More to make video or audio
Other workflows
- MOSS-Image Text to Image Makes a finished image from a one-line idea, including readable English or Chinese text
- Multiple Camera Angles App Re-creates one picture from a new camera angle you set with a 3D camera control
- Qwen Image Edit 2511 (Int8) Edits a photo from a plain instruction, built to keep the person's identity consistent
- Remove Background (BiRefNet) Removes the background from a photo and gives you a transparent cut-out and a mask
- SeedVR2 Image Upscale Enlarges a small or blurry photo into a sharper, higher-resolution image on your computer
- SeedVR2 Video Upscale Turns a grainy, low-resolution video clip into a cleaner, higher-resolution one