NorthernStar 🌟 Your companion for every conversation. NorthernStar is a voice-first AI companion designed to provide empathetic support, encourage positive habit formation, and foster a sense of connection. Built with a focus on creating a safe and supportive space, this app leverages multiple AI services to deliver a natural, human-like conversational experience. Core Features 💬 AI Companion: Engage in open-ended, empathetic conversations through text or real-time voice calls. The AI is designed to be a warm, emotionally intelligent listener. 🎯 Habit Formation: A gamified system to help build and maintain positive daily habits. Track progress, maintain streaks, and earn credits for consistency. 🗣️ Real-time Voice Calls: Experience fluid, live conversations with an ElevenLabs Conversational AI Agent using a real-time audio stream. 📝 Local Speech-to-Text: An offline-first STT server built with Python, Flask, and Vosk for transcribing recorded audio notes. 🫂 Community Connection: A hub to find and connect with other community members on similar journeys. Tech Stack The project contains both the mobile application and a local Python server for STT. Frontend (Mobile App) Framework: React Native with Expo Navigation: Expo Router (File-based) Language: TypeScript UI Components: Custom components with icons from lucide-react-native Backend & AI Services Text AI: OpenAI API (gpt-4o-mini) Conversational Voice AI: ElevenLabs Conversational AI Agent (via WebSockets) Local STT Server: Python, Flask, Vosk Audio Processing: pydub and ffmpeg Getting Started Follow these instructions to get a local copy up and running for development and testing purposes. Prerequisites Node.js (LTS version recommended) Python 3.x and Pip Expo Go app on your mobile device or an Android/iOS emulator Git

1. External Dependencies (Vosk & FFmpeg) This project requires external dependencies that are not included in the repository. Vosk Speech Recognition Model: Download the lightweight English model (vosk-model-small-en-us-0.15) from the Vosk website. Unzip the folder. Rename the folder to model-en and place it inside the /backend directory (you may need to create this directory). FFmpeg: Download the FFmpeg binaries for your operating system from the official FFmpeg website. Place the ffmpeg.exe (or equivalent) binary in a known location on your system. Important: Open the Python server script and update the path to point to your ffmpeg.exe location.
2. Setup Instructions Clone the repository: git clone https://github.com/your-username/NorthernStar.git cd NorthernStar

Configure Environment Variables: Open the app.json file. Locate the extra object. Fill in your secret keys for OPENAI_API_KEY, ELEVENLABS_API_KEY, and ELEVENLABS_AGENT_ID. Install Frontend Dependencies: npm install

Setup and Run the Backend Server: Create a backend folder in the project root if it doesn't exist. Place your Python server script (e.g., STTServerindex.py) inside. Create a virtual environment, activate it, and install the Python dependencies (Flask, vosk, pydub). Run the server: python STTServerindex.py Note: Ensure your computer's firewall allows connections on port 5000. 3. Running the Application With the backend server running in one terminal, start the frontend app in a second terminal. Terminal 2: Start the Frontend App

npx expo start

Scan the QR code with the Expo Go app or run on an emulator. Project Structure NorthernStar/ ├── app/ # Expo Router screens │ ├── (tabs)/ # Main app screens │ │ ├── _layout.tsx │ │ ├── index.tsx # Dashboard │ │ ├── interact.tsx # AI Companion Chat │ │ └── ... │ └── _layout.tsx # Root layout ├── assets/ # Fonts and images ├── services/ # API service logic │ ├── aiService.ts # OpenAI text service │ └── elevenLabsCallService.ts # ElevenLabs voice service └── app.json # App configuration and API keys