MiniMax Music3

Name: MiniMax Music3
Author: MiniMaxAI

ProviderMiniMaxAI

Categorytext-to-audio

LicenseApache-2.0

Downloads8.6K

Stars0

Overview

MiniMax Music3 is a specialized text-to-audio model designed for high-fidelity music generation. For developers, its primary value lies in the ability to transform descriptive prompts into structured audio assets, bridging the gap between conceptual sound design and production-ready output. Unlike general-purpose audio models, Music3 focuses on rhythmic coherence and melodic stability, making it suitable for integration into gaming engines, dynamic content creation tools, or automated soundtracking pipelines. With an Apache-2.0 license, it offers significant flexibility for commercial deployment and local fine-tuning. It competes by balancing computational efficiency with high-sample-rate output, providing a scalable alternative for apps requiring real-time or batch audio synthesis without the overhead of massive proprietary frameworks.

Highlights

High-fidelity text-to-music synthesis for creative workflows
Permissive Apache-2.0 license for commercial integration
Optimized for rhythmic stability and melodic accuracy
Scalable architecture suitable for dynamic audio pipelines

Usage

Install

# Install Hugging Face transformers
pip install transformers torch

SDK Usage

# Load model with transformers
from transformers import AutoModel, AutoTokenizer

model = AutoModel.from_pretrained("MiniMaxAI/MiniMax-Music3")
tokenizer = AutoTokenizer.from_pretrained("MiniMaxAI/MiniMax-Music3")

Hugging Face Download

We recommend downloading the model via the Hugging Face CLI or Hub SDK.

Guidance：Before downloading, install huggingface_hub with:

Guidance

pip install -U huggingface_hub

CLI Download

Download the full repository

huggingface-cli download MiniMaxAI/MiniMax-Music3

Download a single file to a local folder (e.g. config.json into ./dir)

huggingface-cli download MiniMaxAI/MiniMax-Music3 config.json --local-dir ./dir

See the official docs for more CLI options

SDK Download

# 模型下载
from huggingface_hub import snapshot_download
model_dir = snapshot_download('MiniMaxAI/MiniMax-Music3')

Git Download

Make sure git-lfs is installed first

Git Download

git lfs install
git clone https://huggingface.co/MiniMaxAI/MiniMax-Music3

To skip LFS large-file downloads, use:

Skip LFS

GIT_LFS_SKIP_SMUDGE=1 git clone https://huggingface.co/MiniMaxAI/MiniMax-Music3

Model files are hosted on the Hugging Face Hub — download directly via HF CLI / SDK / Git, not through this site.

PyTorch / Transformers Usage

Install Transformers

pip install -U transformers torch

Load the model and run inference

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained('MiniMaxAI/MiniMax-Music3')
tokenizer = AutoTokenizer.from_pretrained('MiniMaxAI/MiniMax-Music3')

Full Documentation

来源: HuggingFace

---
library_name: diffusers
pipeline_tag: text-to-audio
tags:
- music-generation
- text-to-music
- pytorch
- sglang-omni
---

MiniMax Music 3

MiniMax Music 3 is a high-performance music generation model for creating complete songs up to five minutes long. Conditioned on lyrics and a detailed music description, it generates structurally coherent songs with expressive vocals, evolving arrangements, and stable long-form audio quality.

MiniMax Music 3 combines an 8B Global LLM for long-range musical structure, a 0.6B Local LLM for frame-level acoustic detail, and a continuous hidden-state synthesis system based on Flow Matching and Flow-VAE. The model produces 32 kHz, 16-bit stereo WAV audio.

Demo

Explore music generation examples on the MiniMax Music 3 Demo.

Complete Songs with Long-Range Coherence

MiniMax Music 3 natively supports full-song generation up to five minutes. The model maintains musical themes, rhythm, vocal identity, and arrangement progression across long sequences, enabling complete structures such as intro, verse, pre-chorus, chorus, bridge, instrumental break, and outro.

Fine-Grained Music Control

The model accepts two complementary inputs:

Lyrics define the words to be sung and may include explicit section tags such as [Intro], [Verse], [Pre-Chorus], [Chorus], [Post-Chorus], [Bridge], [Instrumental], [Solo], and [Outro].

Music description defines the musical style, emotional progression, vocal performance, instrumentation, arrangement, and production profile.

For precise control, we recommend using a Structured Caption with three sections:

Global Metadata: genre, subgenre, BPM, key, scale, emotional progression, listening scenario, and production profile.

Vocal Details: vocal gender, timbre, performance style, harmony, backing vocals, and vocal effects.

Arrangement: primary and secondary instruments, section-level instrument evolution, groove, bass, percussion, textures, and spatial effects.

This representation allows the model to follow not only a global style, but also the musical development of the song over time.

Hybrid-LM

MiniMax Music 3 uses a hierarchical autoregressive architecture that separates global musical modeling from local acoustic modeling.

The Global LLM (8B) predicts the first RVQ codebook frame by frame and models the song's long-range semantic and structural progression.

The Local LLM (0.6B) predicts the remaining acoustic codebooks within each frame and restores fine-grained acoustic information.

The Global LLM is initialized from Qwen3-8B. During training, its embedding and output layers are first adapted to semantic music tokens. The Global and Local LLMs are then jointly trained to model all RVQ codebooks.

Continuous Hidden-State Synthesis

Instead of decoding only from discrete RVQ tokens, the synthesis module fuses the final hidden states of the Global and Local LLMs. These continuous representations preserve richer acoustic information for vocal articulation, instrumental texture, and temporal continuity.

The synthesis path is:

text

Global and Local LLM hidden states
                ↓
       Hidden-state fusion
                ↓
     Flow Matching (2.4B)
                ↓
        Flow-VAE latent
                ↓
    Flow-VAE Decoder (123M)
                ↓
       32 kHz stereo audio

The Flow-VAE architecture is adapted from MiniMax Speech and retrained for the dynamic range and spectral characteristics of music.

Music Tokenizer

The training tokenizer uses eight layers of Residual Vector Quantization (RVQ):

The first semantic codebook contains 16,384 entries and captures the core musical semantics and structure.

The remaining seven acoustic codebooks contain 1,024 entries each and represent residual acoustic details.

Training first optimizes the semantic codebook, then jointly trains all eight codebooks. At inference time, waveform synthesis uses the fused LLM hidden states and does not require the discrete tokenizer decoder.

How to Use

MiniMax Music 3 is supported by SGLang-Omni. Follow the official installation guide to prepare the runtime environment.

Download the Model

bash

hf download MiniMaxAI/MiniMax-Music3 --local-dir /path/to/minimax_ttm

We recommend the following inference frameworks to serve the model:

SGLang \- see cookbook

diffusers \- see diffusers docs

ComfyUI see comfyUI tutorials

Serve with SGLang-Omni

bash

sgl-omni serve --model-path MiniMaxAI/MiniMax-Music3 --port 8000

Generate Music

The service uses the shared speech API. Put the lyrics in input and the music description in instructions. Put lyric structure tags such as [Verse] and [Chorus] on their own lines.

bash

curl http://127.0.0.1:8000/v1/audio/speech \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "MiniMaxAI/MiniMax-Music3",
    "input": "[Verse]\nMorning light filtering through the pine\n[Chorus]\nSoftly the world begins to breathe",
    "instructions": "A warm acoustic pop song with intimate female vocals, fingerpicked guitar, soft piano, and a gradual emotional build into a wide final chorus.",
    "response_format": "wav",
    "seed": 7,
    "max_new_tokens": 750,
    "stream": false
  }' \
  --output minimax_music3.wav

max_new_tokens sets the maximum number of audio frames at 25 frames per second. Generation may finish before this limit when the model emits an end-of-audio token. The response is a 32 kHz, 16-bit stereo WAV file.

R

View official source