---
title: ACE-Step1.5-MLX-4bit
canonical_url: "https://www.modelscope.cn/models/mlx-community/ACE-Step1.5-MLX-4bit"
md_url: "https://www.modelscope.cn/models/mlx-community/ACE-Step1.5-MLX-4bit.md"
repository: mlx-community/ACE-Step1.5-MLX-4bit
last_updated: 2026-04-14
model_type:
  - acestep
architectures:
  - AceStepConditionGenerationModel
base_model:
  - ACE-Step/ACE-Step1.5
base_model_relation: finetune
parameters: 1.3B
tensor_type:
  - F32
  - U32
library_name:
  - mlx
  - safetensors
  - diffusers
  - pytorch
frameworks:
  - pytorch
downloads: 50
stars: 0
tags:
  - mlx
  - music-generation
  - ace-step
  - audio
  - text-to-music
---

# ACE-Step1.5-MLX-4bit

> ACE-Step1.5-MLX-4bit - mlx-community 在 ModelScope 开源的模型。ACE-Step 1.5 MLX (4-bit Quantized)

mlx-community/ACE-Step1.5-MLX-4bit 是 ModelScope 魔搭社区上的 1.3B 参数机器学习模型，基于 ACE-Step/ACE-Step1.5 构建。

- **Repository**: mlx-community/ACE-Step1.5-MLX-4bit
- **Parameters**: 1.3B
- **Base model**: ACE-Step/ACE-Step1.5
- **Tags**: mlx, music-generation, ace-step, audio, text-to-music
- **Downloads**: 50
- **Stars**: 0
- **Last updated**: 2026-04-14

Source: https://www.modelscope.cn/models/mlx-community/ACE-Step1.5-MLX-4bit

---

# ACE-Step 1.5 MLX (4-bit Quantized)

4-bit quantized MLX weights for [ACE-Step/ACE-Step1.5](https://huggingface.co/ACE-Step/ACE-Step1.5).

- Decoder and encoder quantized to 4-bit (group_size=64)
- VAE, tokenizer, and detokenizer kept in full precision
- **2.2GB** main model + 0.7GB VAE + 2.4GB text encoder

## Usage

```python
from mlx_audio.tts import load

model = load("mlx-community/ACE-Step1.5-MLX-4bit")

for result in model.generate(
    text="upbeat electronic dance music with energetic synthesizers",
    duration=30.0,
):
    audio = result.audio  # [samples, 2] stereo @ 48kHz
    sample_rate = result.sample_rate
```

## With Vocals

```python
for result in model.generate(
    text="English pop song with clear female vocals, catchy melody",
    lyrics="""[verse]
Dance with me tonight
Under the neon lights

[chorus]
We're alive, we're on fire
Dancing higher and higher
""",
    duration=60.0,
    vocal_language="en",
):
    ...
```

The model uses a 5Hz Language Model planner by default (`use_lm=True`) which generates
a song blueprint before running the diffusion transformer.