File size: 1,226 Bytes
f8b2e54 | 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 34 35 36 37 38 39 | ---
language: ar
tags:
- text-to-speech
- egyptian-arabic
- voice-cloning
- vibevoice
---
# VibeVoice Egyptian Arabic — cfg_scale=3.5
Fine-tuned VibeVoice on Egyptian Arabic dialect (checkpoint-9160).
## cfg_scale=3.5
Natural prosody, moderate voice cloning. Good for conversational TTS.
_Lowest safe value — values below 3.0 cause over-generation past script end._
## Repo contents
| File | Description |
|---|---|
| `model.safetensors` | Merged model weights (5.1 GB, single shard) |
| `config.json` | Model architecture config |
| `tokenizer.json` + friends | Qwen2.5 tokenizer files |
| `preprocessor_config.json` | Audio processor settings |
| `voices/egyptian_male.wav` | Reference voice for male speaker |
| `voices/egyptian_female.wav` | Reference voice for female speaker |
| `samples/demo_cfg3.5.wav` | Sample output at this cfg_scale |
## Backend usage
```python
from huggingface_hub import hf_hub_download, snapshot_download
snapshot_download("MohammedEhab20/vibe-voice-egyptian-cfg35", local_dir="./model")
male_voice = "./model/voices/egyptian_male.wav"
female_voice = "./model/voices/egyptian_female.wav"
# Run inference:
# python inference_from_file.py --model_path ./model --cfg_scale 3.5 ...
```
|