F5-TTS: Fine-Tuned Arabic Speech Synthesis Model

License: CC BY-NC 4.0
Base Model: SWivid/F5-TTS

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, a crowdsourced dataset featuring diverse Arabic accents and dialects. Additional datasets may be incorporated in future updates to improve dialectal coverage and pronunciation accuracy.

Model Information

Base Model: SWivid/F5-TTS
Current Status: Ongoing fine-tuning (Temporary Checkpoints Available)
Training Configuration:
- Batch Size: TBD
- Max Samples: TBD
- Training Steps: TBD
- (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

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

Replace the Model File
- Navigate to the displayed file location.
- Rename the existing model file:
```
model_1200000.safetensors → model_1200000.safetensors.bak
```
- Download the Arabic checkpoint and vocabulary files from this repository and place them in the same directory.
Restart the Application
- Relaunch the F5-TTS application to load the Arabic fine-tuned model.

Alternative Methods

GitHub Repository: Follow the F5-TTS setup instructions, 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! 🚀