Instructions to use commandeaw/OmniVoice-MLX-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use commandeaw/OmniVoice-MLX-4bit with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir OmniVoice-MLX-4bit commandeaw/OmniVoice-MLX-4bit
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
File size: 719 Bytes
28ec8a1 6c201a1 28ec8a1 a267cf7 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 | ---
license: apache-2.0
base_model: k2-fsa/OmniVoice
base_model_relation: quantized
library_name: mlx-audio
pipeline_tag: text-to-speech
tags:
- mlx
---
OmniVoice in MLX (4bit). For [mlx-audio](https://github.com/Blaizzy/mlx-audio) on Apple Silicon.
## Usage
```bash
pip install mlx-audio
python -m mlx_audio.tts.generate \
--model commandeaw/OmniVoice-MLX-4bit \
--text "สวัสดีค่ะ ยินดีต้อนรับ" \
--lang_code th \
--ref_audio reference.wav
```
`--ref_audio` is optional (zero-shot voice cloning); drop it for the default voice.
## Size
| variant | repo | vs base (fp32) |
|---|---|---|
| bf16 | 1.6 GB | -50% |
| 8bit | 1.1 GB | -68% |
| 4bit | 0.75 GB | -77% |
|