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
Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -7,4 +7,25 @@ tags:
|
|
| 7 |
- mlx
|
| 8 |
---
|
| 9 |
|
| 10 |
-
OmniVoice in MLX (4bit). For mlx-audio.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
- mlx
|
| 8 |
---
|
| 9 |
|
| 10 |
+
OmniVoice in MLX (4bit). For [mlx-audio](https://github.com/Blaizzy/mlx-audio) on Apple Silicon.
|
| 11 |
+
|
| 12 |
+
## Usage
|
| 13 |
+
|
| 14 |
+
```bash
|
| 15 |
+
pip install mlx-audio
|
| 16 |
+
python -m mlx_audio.tts.generate \
|
| 17 |
+
--model commandeaw/OmniVoice-MLX-4bit \
|
| 18 |
+
--text "สวัสดีค่ะ ยินดีต้อนรับ" \
|
| 19 |
+
--lang_code th \
|
| 20 |
+
--ref_audio reference.wav
|
| 21 |
+
```
|
| 22 |
+
|
| 23 |
+
`--ref_audio` is optional (zero-shot voice cloning); drop it for the default voice.
|
| 24 |
+
|
| 25 |
+
## Size
|
| 26 |
+
|
| 27 |
+
| variant | repo | vs base (fp32) |
|
| 28 |
+
|---|---|---|
|
| 29 |
+
| bf16 | 1.6 GB | -50% |
|
| 30 |
+
| 8bit | 1.1 GB | -68% |
|
| 31 |
+
| 4bit | 0.75 GB | -77% |
|