---
title: CosyVoice2-0.5B-8bit
canonical_url: "https://www.modelscope.cn/models/mlx-community/CosyVoice2-0.5B-8bit"
md_url: "https://www.modelscope.cn/models/mlx-community/CosyVoice2-0.5B-8bit.md"
repository: mlx-community/CosyVoice2-0.5B-8bit
last_updated: 2025-12-12
pipeline_tag: text-to-speech
tasks:
  - text-to-speech
model_type:
  - cosyvoice2
base_model:
  - FunAudioLLM/CosyVoice2-0.5B
base_model_relation: finetune
parameters: 388.9M
tensor_type:
  - F16
  - U32
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - en
  - zh
  - ja
  - ko
downloads: 125
stars: 1
tags:
  - mlx
  - tts
  - cosyvoice2
---

# CosyVoice2-0.5B-8bit

> CosyVoice2-0.5B-8bit - mlx-community 在 ModelScope 开源的模型。mlx-community/CosyVoice2-0.5B-8bit

mlx-community/CosyVoice2-0.5B-8bit 是 ModelScope 魔搭社区上的 388.9M 参数text-to-speech模型，基于 FunAudioLLM/CosyVoice2-0.5B 构建。

- **Repository**: mlx-community/CosyVoice2-0.5B-8bit
- **Tasks**: text-to-speech
- **Parameters**: 388.9M
- **Base model**: FunAudioLLM/CosyVoice2-0.5B
- **Tags**: mlx, tts, cosyvoice2
- **Downloads**: 125
- **Stars**: 1
- **Last updated**: 2025-12-12

Source: https://www.modelscope.cn/models/mlx-community/CosyVoice2-0.5B-8bit

---

# mlx-community/CosyVoice2-0.5B-8bit

This model was converted to MLX format from [FunAudioLLM/CosyVoice2-0.5B](https://huggingface.co/FunAudioLLM/CosyVoice2-0.5B) using [mlx-audio-plus](https://github.com/DePasqualeOrg/mlx-audio-plus) version **0.1.2**.

## Usage

```bash
pip install -U mlx-audio-plus
```

### Inference Modes

| Mode | Parameters | Description |
|------|------------|-------------|
| Cross-lingual | `ref_audio` | Zero-shot TTS (default) |
| Zero-shot | `ref_audio` + `ref_text` | Better quality with transcription |
| Instruct | `ref_audio` + `instruct_text` | Style control (e.g., "speak slowly") |
| Voice Conversion | `source_audio` + `ref_audio` | Convert audio to target voice |

### Command line

```bash
# Cross-lingual (default)
mlx_audio.tts --model mlx-community/CosyVoice2-0.5B-8bit --text "Hello!" --ref_audio ref.wav

# Zero-shot (with transcription)
mlx_audio.tts --model mlx-community/CosyVoice2-0.5B-8bit --text "Hello!" --ref_audio ref.wav --ref_text "Transcription of ref audio."

# Instruct (style control)
mlx_audio.tts --model mlx-community/CosyVoice2-0.5B-8bit --text "Hello!" --ref_audio ref.wav --instruct_text "Speak slowly and calmly"

# Voice Conversion
mlx_audio.tts --model mlx-community/CosyVoice2-0.5B-8bit --source_audio source.wav --ref_audio ref.wav
```

### Python

```python
from mlx_audio.tts.generate import generate_audio

generate_audio(
    text="Hello, this is CosyVoice2 on MLX!",
    model="mlx-community/CosyVoice2-0.5B-8bit",
    ref_audio="reference.wav",
    file_prefix="output",
)
```