Orator TTS Engine

High-quality neural text-to-speech with 50+ voices across 9 languages

Click for Demo ☝️ [Will update the demo]

UPDATE: Install the python package using pip install -e .

✨ Features

• 🌍 9 Languages: American English, British English, Spanish, French, Hindi, Italian, Portuguese, Japanese, Chinese
• 🎭 55 Voices: Male and female voices with unique personalities
• ⚡ Lightning Fast: GPU-accelerated inference with streaming audio
• 🎯 macOS Hotkey: Double-tap Option key ⌥ for instant TTS anywhere
• 🔊 High Quality: Super high quality neural audio synthesis
• 🚀 Easy Setup: Installation through UV package manager
• 📱 System-Wide: Works with any macOS application

🗣️ Available Voices & Languages

Voices/Languages Available

🇺🇸 American English (a)

Female Voices:

• af (default)
• af_alloy
• af_aoede
• af_bella
• af_heart
• af_jessica
• af_kore
• af_nicole
• af_nova
• af_river
• af_sarah
• af_sky

Male Voices:

• am_adam
• am_echo
• am_eric
• am_fenrir
• am_liam
• am_michael
• am_onyx
• am_puck
• am_santa

🇬🇧 British English (b)

Female Voices:

• bf_alice
• bf_emma
• bf_isabella
• bf_lily

Male Voices:

• bm_daniel
• bm_fable
• bm_george
• bm_lewis

🇪🇸 Spanish (e)

Female Voices:

• ef_dora

Male Voices:

• em_alex
• em_santa

🇫🇷 French (f)

Female Voices:

• ff_siwis

🇮🇳 Hindi (h)

Female Voices:

• hf_alpha
• hf_beta

Male Voices:

• hm_omega
• hm_psi

🇮🇹 Italian (i)

Female Voices:

• if_sara

Male Voices:

• im_nicola

🇯🇵 Japanese (j)

Female Voices:

• jf_alpha
• jf_gongitsune
• jf_nezumi
• jf_tebukuro

Male Voices:

• jm_kumo

🇧🇷 Portuguese (p)

Female Voices:

• pf_dora

Male Voices:

• pm_alex
• pm_santa

🇨🇳 Chinese (z)

Female Voices:

• zf_xiaobei
• zf_xiaoni
• zf_xiaoxiao
• zf_xiaoyi

Male Voices:

• zm_yunjian
• zm_yunxi
• zm_yunxia
• zm_yunyang

🚀 Quick Start

Why UV? The Future of Python Package Management

We recommend UV for this project because it's:

• ⚡ 10-100x faster than pip
• 🔒 More secure with built-in dependency resolution
• 🎯 Zero configuration - works out of the box
• 🔄 Drop-in replacement for pip/pipenv/poetry
• 🌟 Industry standard - used by major Python projects

Installation

Option A: Install as editable package (Recommended)

1. Clone and install the package:

   # Clone repo
   cd Orator
   
   # Create virtual environment
   python3 -m venv .venv
   source .venv/bin/activate  # On macOS/Linux
   
   # Install in editable mode
   pip install -e .

Option B: Using UV (Fast alternative)

1. Install UV (if you don't have it):

• Assumed that python is already installed on your system.

  pip install uv 

2. Clone and setup the project:

   # Clone repo
   cd Orator
   
   # Create virtual environment and install dependencies
   uv venv --python=3.11
   source .venv/bin/activate  # On macOS/Linux
   uv pip install -r requirements.txt

Additional Setup (Required for both options)

3. Install espeak-ng (required for phonemization):

   # macOS
   brew install espeak-ng
   
   # Verify installation
   espeak-ng --version
   
   #eSpeak NG text-to-speech: 1.51  Data at: /opt/homebrew/Cellar/espeak-ng/1.51/share/espeak-ng-data

4. Download model and voices (if not included):

   uv pip install -U "huggingface_hub[cli]"
   
   # Download model
   huggingface-cli download hexgrad/Kokoro-82M --include "onnx/model.onnx" --local-dir ./kokoro_model_onnx/
   
   # Download voices
   huggingface-cli download onnx-community/Kokoro-82M-v1.0-ONNX --include "voices/*" --local-dir ./kokoro_model_onnx/voices

5. Language Pack

• By default "en-core-web-sm" is installed through requirements for English, navigate and install other small language packs from spaCy.

🎯 Usage

1. macOS Hotkey Application

Grant Accessibility Permissions First:

1. Open System Preferences → Security & Privacy → Privacy
2. Select "Accessibility" from the left panel
3. Click the lock icon and enter your password
4. Add your terminal application (Terminal.app, iTerm2, etc.)
5. Ensure it's checked/enabled

Run the hotkey application:

# Make sure your are inside the virtual environment
python3 macos_tts_hotkey.py

How to use:

• Select any text in any macOS application
• Double-tap the Option key (⌥) quickly to start TTS
• Press Escape key to stop TTS playback at any time
• Listen to the text being read aloud!

⚙️ Configuration

Hotkey Application Config

Edit config_hotkey.json:

{
    "model_path": "kokoro-v1_0.pth",
    "voices_dir": "voices",
    "voice": "af_bella",
    "speed": 1.0,
    "device": "auto"
}

Voice Selection

Choose voices by language prefix:

• af_* / am_* - American English
• bf_* / bm_* - British English
• ef_* / em_* - Spanish
• ff_* - French
• hf_* / hm_* - Hindi
• if_* / im_* - Italian
• jf_* / jm_* - Japanese
• pf_* / pm_* - Portuguese
• zf_* / zm_* - Chinese

🛠️ Troubleshooting

Common Issues

"Failed to start keyboard monitoring"

• Grant Accessibility permissions in System Preferences
• Restart the application after granting permissions

"espeak-ng not found"

# Install espeak-ng
brew install espeak-ng

# Verify installation
which espeak-ng

"Model file not found"

• Ensure kokoro-v1_0.onnx is in the kokoro_model_onnx directory
• Check file permissions and path

"CUDA out of memory"

# Use CPU instead
config.device = "cpu"

# Or reduce batch size for long texts

"Voice file not found"

• Ensure voice files are in the voices/ directory
• Check that the voice name matches exactly (case-sensitive)

"Stop functionality not working"

• Ensure the application has focus or accessibility permissions
• Try pressing Escape key while TTS is actively playing
• Check terminal logs for any error messages

Performance Optimization

• GPU Usage: Automatic CUDA detection, falls back to CPU
• Memory Management: Automatic cleanup after generation
• Streaming: Use generate_audio_stream() for long texts
• Caching: Voice packs are cached after first load

🤝 Contributing

We welcome contributions! Please feel free to:

• Report bugs and issues
• Suggest new features
• Submit pull requests
• Add new voice packs
• Improve documentation

🗺️ Roadmap

• Streaming Audio chunks for Long Formers (Controlled low latency)
• Speed Controls for Audio Stream
• LLM driven Agentic AI Capabilities
• Native MacOS application/interface for UI driven audio controlls
• UI voice swap controlls

---

🤝 Get Involved

Want to help shape the future of Kokoro TTS? Here's how:

• 🐛 Report Issues - Help us identify bugs and improvements
• 💡 Suggest Features - Share your ideas for new functionality
• 🔧 Contribute Code - Submit PRs for features or fixes
• 🎨 Design UI/UX - Help design the native app interface
• 📝 Write Documentation - Improve guides and tutorials
• 🗣️ Add Voices - Contribute new voice packs and languages

🙏 Acknowledgments

• Built on the amazing Kokoro TTS model
• Powered by ONNX and modern neural architectures
• Inspired by the need for accessible, high-quality TTS

---

📊 Repository Stats

---

Made with ❤️ for the open-source community