---
title: VibeVoice7b-low-vram
canonical_url: "https://www.modelscope.cn/models/AI-ModelScope/VibeVoice7b-low-vram"
md_url: "https://www.modelscope.cn/models/AI-ModelScope/VibeVoice7b-low-vram.md"
repository: AI-ModelScope/VibeVoice7b-low-vram
last_updated: 2025-10-24
license: mit
pipeline_tag: text-to-speech
tasks:
  - text-to-speech
parameters: 5.4B
tensor_type:
  - F32
  - U8
  - BF16
library_name:
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - en
downloads: 177
stars: 0
tags:
  - text-to-speech
  - speech-synthesis
  - quantized
  - low-vram
  - vibevoice
---

# VibeVoice7b-low-vram

> VibeVoice7b-low-vram - AI-ModelScope 在 ModelScope 开源的模型。VibeVoice 7B - Low VRAM Quantized Models

AI-ModelScope/VibeVoice7b-low-vram 是 ModelScope 魔搭社区上的 5.4B 参数text-to-speech模型，采用 mit 许可。

- **Repository**: AI-ModelScope/VibeVoice7b-low-vram
- **License**: mit
- **Tasks**: text-to-speech
- **Parameters**: 5.4B
- **Tags**: text-to-speech, speech-synthesis, quantized, low-vram, vibevoice
- **Downloads**: 177
- **Stars**: 0
- **Last updated**: 2025-10-24

Source: https://www.modelscope.cn/models/AI-ModelScope/VibeVoice7b-low-vram

---

# VibeVoice 7B - Low VRAM Quantized Models

Pre-quantized versions of VibeVoice 7B for low VRAM GPUs.

## Available Versions

- **4bit/** - 4-bit quantized model (~8GB VRAM needed)
- **8bit/** - 8-bit quantized model (~12GB VRAM needed) - NOTE: Removed 8 bit until I can test it again. I'll re-up it soon.

## Usage

```python
from vibevoice.modular.modeling_vibevoice_inference import VibeVoiceForConditionalGenerationInference
from vibevoice.processor.vibevoice_processor import VibeVoiceProcessor

# For 4-bit model
model = VibeVoiceForConditionalGenerationInference.from_pretrained(
    "Dannidee/VibeVoice7b-low-vram/4bit",
    device_map='cuda',
    torch_dtype=torch.bfloat16,
)
processor = VibeVoiceProcessor.from_pretrained("Dannidee/VibeVoice7b-low-vram/4bit")
```

## VRAM Requirements

- **4-bit**: ~8 GB total VRAM
- **8-bit**: ~12 GB total VRAM
- **Original**: ~19 GB total VRAM

See individual model folders for detailed information.