---
datasets:
- mozilla-foundation/common_voice_17_0
language:
- ar
base_model:
- SWivid/F5-TTS
pipeline_tag: text-to-speech
tags:
- speech
- f5-tts
- arabic
---
# F5-TTS: Fine-Tuned Arabic Speech Synthesis Model

## Overview
This project fine-tunes the F5-TTS model for high-quality Arabic speech synthesis, incorporating regional diversity in pronunciation and accents. The fine-tuning process is ongoing, and temporary checkpoints are provided as progress updates. Future iterations will include improved models with enhanced accuracy and naturalness.

## License
This model is released under the **CC BY-NC 4.0** license, which allows free usage, modification, and distribution for **non-commercial** purposes.

## Datasets
Training is based on the **Common Voice Arabic Dataset** so basically the model support MSA
## Model Information
- **Base Model:** SWivid/F5-TTS  
- **Current Status:** Ongoing fine-tuning (Temporary Checkpoints Available)  
- *(Final training parameters will be updated upon completion of fine-tuning.)*

## Usage Instructions
To use the fine-tuned Arabic model, follow these steps:

### Method 1: Manual Model Replacement
1. **Run the F5-TTS Application**  
   - Start the application and locate the model file path displayed in the terminal. Example:  
     ```
     model : C:\Users\yourname\.cache\huggingface\hub\models--SWivid--F5-TTS\snapshots\995ff41929c08ff968786b448a384330438b5cb6\F5TTS_Base\model_1200000.safetensors
     ```
2. **Replace the Model File**  
   - Download the **Arabic checkpoint** and **vocabulary files** from this repository and place them in the same directory.  
3. **Restart the Application**  
   - Relaunch the F5-TTS application to load the Arabic fine-tuned model.
    - Download the **Arabic checkpoint** and **vocabulary files** from this repository and use them instead of the basecheckpoint.

### Alternative Methods
- **GitHub Repository:** Follow the [F5-TTS setup instructions](https://github.com/SWivid/F5-TTS), but replace the default model with the Arabic checkpoint and vocabulary files provided here.

## Contributions & Collaboration
This model is a **work in progress**, and community contributions are highly encouraged! Suggestions, improvements, and dataset contributions are welcome to refine its performance across different Arabic dialects.

### Recommendations for Better Results
- Use **clear reference audio** with minimal background noise.  
- Ensure **balanced audio levels** for improved synthesis quality.  
- Contributions in **dataset expansion** and **model evaluation** are highly valuable.

If you have any questions or suggestions, feel free to reach out! 🚀