--- title: Arabic F5-TTS emoji: 🎙️ colorFrom: green colorTo: blue sdk: gradio sdk_version: 4.44.0 app_file: app.py pinned: false license: mit models: - IbrahimSalah/Arabic-F5-TTS-v2 tags: - text-to-speech - tts - arabic - voice-cloning - f5-tts --- # 🎙️ Arabic Text-to-Speech (F5-TTS Model) High-quality Arabic text-to-speech synthesis using the F5-TTS model with voice cloning capabilities. ## 🌟 Features - **Voice Cloning**: Upload a reference audio to clone the voice style - **Diacritized Text Support**: Uses fully diacritized Arabic text (تشكيل) for accurate pronunciation - **High Quality**: Natural-sounding speech with controllable parameters - **Fast Generation**: Efficient inference with NFE steps control - **Speed Control**: Adjust speech speed from 0.5x to 2.0x ## 🚀 Quick Start 1. **Enter diacritized Arabic text** (with تشكيل) 2. **Use the default reference audio** or upload your own (WAV format, 5-30 seconds) 3. **Provide the diacritized transcript** of your reference audio 4. **Adjust settings** (optional) - NFE steps, CFG strength, speed 5. **Click "Generate Speech"** ## ⚠️ Important: Diacritized Text Required This model requires **fully diacritized Arabic text (تشكيل)** for both: - Input text to synthesize - Reference audio transcript ### How to Add Diacritics: **Option 1: Use AI (Recommended)** - Ask ChatGPT, Claude, or Gemini: "أضف التشكيل الكامل للنص التالي: [your text]" **Option 2: Online Tools** - [Mishkal Tashkeel](https://tahadz.com/mishkal) - [Harakat.ai](https://harakat.ai) ## 🎯 Model Information - **Model ID**: `IbrahimSalah/Arabic-F5-TTS-v2` - **Language**: Modern Standard Arabic (MSA) and dialects - **Sample Rate**: 24kHz - **Architecture**: Flow Matching based TTS (F5-TTS) ## 🔧 Advanced Settings - **NFE Steps**: Number of function evaluations (16-64, default: 32) - Higher = better quality but slower - **CFG Strength**: Classifier-free guidance strength (0-3, default: 1.8) - Controls adherence to prompt - **Speed**: Playback speed (0.5-2.0, default: 1.0) ## 🔗 Related Resources - **Model Card**: [IbrahimSalah/Arabic-F5-TTS-v2](https://huggingface.co/IbrahimSalah/Arabic-F5-TTS-v2) - **Spark TTS Arabic**: [IbrahimSalah/Arabic-TTS-Spark](https://huggingface.co/IbrahimSalah/Arabic-TTS-Spark) - **Report Issues**: [Discussions](https://huggingface.co/IbrahimSalah/Arabic-F5-TTS-v2/discussions) ## 📄 License MIT License