Local TTS - Kokoro

A powerful, locally-run Text-to-Speech (TTS) application built with the Kokoro TTS engine. This desktop application provides a modern GUI for generating natural-sounding speech in multiple languages and accents without requiring cloud services.

Features

✨ Multi-Language Support

• American English, British English, Indian English, and more
• Chinese (Mandarin), French, German, Italian, Japanese, Spanish, and Polish
• 50+ diverse voice options including multiple speakers per language

🎙️ Diverse Voice Library

• Gender-specific voices (male/female indicators)
• Multiple accents and personality traits per language
• Native speaker quality for each supported language

🔧 Advanced Controls

• Adjustable speech rate and pitch
• Real-time text input with live preview
• Batch processing for multiple sentences
• Audio export in WAV format

💻 User-Friendly Interface

• Modern desktop GUI built with Tkinter
• Light/Dark theme support
• Keyboard shortcuts (Ctrl+Enter to generate, Esc to cancel)
• Progress tracking and real-time generation status

🚀 Performance Optimized

• GPU acceleration support (CUDA/Metal)
• Lazy device initialization for fast startup
• Per-language pipeline caching
• Multi-threaded audio generation

Installation

⚡ Quick Start (Automated)

Windows:

• Double-click install.bat
• Wait 5-10 minutes
• Double-click run_app.bat

macOS/Linux:

chmod +x install.sh && ./install.sh
source .venv/bin/activate
python app.py

Manual Setup

Requirements: Python 3.10+ | 2GB disk | 500MB download

Windows:

python -m venv .venv
.venv\Scripts\activate
python -m pip install torch==2.1.2 torchaudio==2.1.2 torchvision==0.16.2 --index-url https://download.pytorch.org/whl/cpu
python -m pip install -r requirements.txt
python app.py

macOS/Linux:

python3 -m venv .venv
source .venv/bin/activate
python -m pip install torch==2.1.2 torchaudio==2.1.2 torchvision==0.16.2
python -m pip install -r requirements.txt
python app.py

Usage

Basic Text-to-Speech Generation

1. Select a Voice: Choose from the dropdown menu. Voices are grouped by language and labeled with gender indicators (👩 for female, 👨 for male).

2. Enter Text: Type or paste the text you want to convert to speech in the input area.

3. Adjust Settings:

   • Speech Rate: Control how fast the speech is generated (0.5 - 2.0x)
   • Pitch: Adjust voice pitch characteristics

4. Generate: Click the "Generate" button or press Ctrl+Enter

5. Listen & Export: Play the audio preview and save as WAV file

Keyboard Shortcuts

• Ctrl+Enter: Start generation
• Esc: Cancel ongoing generation
• Ctrl+O: Open file dialog to save audio

Output Formats

Generated audio files are saved as WAV format at 24kHz sample rate for optimal quality.

Supported Languages & Voices

American English (af*, am*)

11 female voices: Alloy, Aoede, Bella, Heart, Jessica, Kore, Nicole, Nova, River, Sarah, Sky 10 male voices: Adam, Echo, Eric, Fenrir, Liam, Michael, Onyx, Puck, Santa

British English (bf*, bm*)

4 female voices: Alice, Emma, Isabella, Lily 4 male voices: Daniel, Fable, George, Lewis

Additional Languages

• Indian English: ef_dora, em_alex, em_santa
• French: ff_siwis
• German: hf_alpha, hf_beta, hm_omega, hm_psi
• Italian: if_sara, im_nicola
• Japanese: jf_alpha, jf_gongitsune, jf_nezumi, jf_tebukuro, jm_kumo
• Portuguese: pf_dora, pm_alex, pm_santa
• Chinese (Mandarin): zf_xiaobei, zf_xiaoni, zf_xiaoxiao, zf_xiaoyi

Requirements

See requirements.txt for the complete dependency list. Key dependencies include:

• Kokoro: Core TTS engine
• PyTorch: Deep learning framework with GPU support
• Tkinter: GUI framework
• SoundFile: Audio file I/O
• NumPy: Numerical computing
• spaCy: NLP for text processing
• TTS ecosystem: espeak-ng, g2p-en, and other speech synthesis tools

Architecture

• app.py: Main application with GUI and TTS pipeline management
• assets/: Application resources (icon, images)
• kokoro: TTS engine and voice synthesis models

The application uses:

• Thread-based audio generation to keep the GUI responsive
• Queue-based communication between generation and UI threads
• Lazy device (GPU/CPU) initialization for startup performance
• Pipeline caching to avoid reloading models when switching voices

Performance Tips

• First Generation: May take longer as models are loaded into VRAM
• GPU Acceleration: Ensure CUDA/Metal is properly installed for ~3-5x speedup
• Batch Processing: Generate multiple sentences at once for efficiency
• Memory: Close other applications to maximize available VRAM for larger models

Troubleshooting

Models not found

Ensure model files are properly installed with the Kokoro package. The application will attempt to download them automatically on first run.

Slow generation

• Check if GPU acceleration is enabled: look for CUDA/Metal in the generation log
• Close unnecessary applications to free up memory
• Consider reducing speech rate for faster processing

Audio quality issues

• Ensure you're using a supported voice for your language
• Try adjusting pitch and rate settings
• Regenerate with the same settings to ensure consistency

Installation issues

If you encounter pip errors during installation:

pip install --upgrade pip setuptools wheel
pip install -r requirements.txt

License & Credits

Built with the Kokoro TTS engine - a state-of-the-art multilingual text-to-speech synthesis system.

This application is provided as-is. Please refer to the licenses of individual dependencies for usage rights.

Contributing

Contributions are welcome! Feel free to:

• Report bugs and issues
• Suggest new features
• Improve documentation
• Optimize performance

Support

For issues specific to Kokoro TTS, visit the Kokoro GitHub repository.

---

Note: This is a local-only application. All processing happens on your machine - no data is sent to external services.