Spritely AI 🧚🏼‍♀️ ✨ 🎙️

Free humanity from the keyboard - Your AI-powered voice companion

Spritely AI is a powerful desktop application that enables real-time audio transcription with AI analysis. It combines local audio processing with cloud-based AI to provide a seamless voice-to-text experience.

🌟 Features

Desktop App

• Real-time audio transcription using Deepgram's Nova-2 model
• Multiple transcription modes:
  • Direct field input (Cmd+Alt+L)
  • AI-analyzed input (Cmd+Alt+K)
• Speaker diarization support
• System-wide keyboard shortcuts
• Local audio processing for privacy
• Automatic microphone selection and configuration

🗺️ Roadmap

• Always-on listening mode with wake word detection. See local STT.
• Create a md file for database. Use an LLM instead of vector similarity search. Code example: text
• Add meeting summaries to md file.
• Polish the tkinter UI, i.e for meeting summary and transcription.
• Add Greptile API to tools.

Bug Fixes

• Spritely's spoken output cuts off without completing the LLMs entire response

🚀 Getting Started

Prerequisites

• Python 3.12+
• MacOS (Windows support coming soon)
• API keys for the following services
  • Elevenlabs account
  • Deepgram
  • Groq
  • Anthropic
• You will need to give keystroke permissions to the app for the shortcuts

Desktop App Setup

1. Clone the repository:

git clone https://github.com/spritelyai/spritely-ai.git

2. Install dependencies:

cd spritely-ai
python3.12 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
touch .env
# add your api keys to .env

3. Run the app:

python main.py

🎯 Usage

Desktop Shortcuts

• Cmd+Alt+K: Start/stop AI-analyzed transcription
• Cmd+Alt+L: Start/stop direct field transcription
• ESC: Stop current transcription

Permissions

The app requires:

• Microphone access
• Accessibility permissions (for keyboard shortcuts)
• Internet connection (for AI analysis)

🏗️ Architecture

Desktop App Components

• Audio Capture: PyAudio
• Transcription: Deepgram SDK
• Voice Synthesis: Cartesia
• Keyboard Control: pynput

🤝 Contributing

We welcome contributions! Please see our Contributing Guide for details.

1. Fork the repository
2. Create your feature branch
3. Commit your changes
4. Push to the branch
5. Open a Pull Request

📝 License

This project is licensed under the AGPL-3.0 for non-commercial use.

Commercial Use For commercial use or deployments requiring a setup fee, please contact us for a commercial license at michael@flowon.ai.

By using this software, you agree to the terms of the license.

🙏 Acknowledgments

• Deepgram for real-time transcription
• Cartesia for voice synthesis
• All our contributors and supporters