CrySense 👶🎙️

AI-Powered Baby Cry Analysis & Wellness Platform

CrySense is a production-grade, clinical-style SaaS platform designed to analyze infant cries using deep learning, helping parents identify emotional states (e.g., hunger, pain, discomfort, tiredness) and track wellness trends over time. Built with a React/Vite frontend, FastAPI backend, and Supabase database.

---

🚀 Key Features

• Clinical-Grade AI Classifier: Uses a custom Convolutional Neural Network (CNN) trained on Mel-frequency cepstral coefficients (MFCCs) to classify infant cry signatures.
• Real-time Waveform Recording: Securely capture and analyze 5-second audio recordings directly in the browser with visual feedback.
• Comprehensive Wellness Dashboard: Track historical trends, average confidence levels, and common cry reasons.
• Production Auth System: Fully integrated email/password and Google OAuth authentication flows using Supabase PKCE flow.
• Modern Design System: Sleek clinical-teal theme (#1B6B8A) with responsive layouts, fluid animations, and a seamless Dark Mode toggle.
• Privacy First: Waveform features are analyzed on-the-fly and classification metadata is stored securely in your database without storing raw audio files.

---

📂 Project Architecture

CrySense/
├── frontend/               # React (Vite) Single Page Application
│   ├── src/
│   │   ├── components/     # Reusable UI components (Navbar, Waveform recorder, etc.)
│   │   ├── context/        # React Auth Context (Supabase state)
│   │   ├── hooks/          # Custom hooks (predictions, audio recorder, dark mode)
│   │   ├── pages/          # App screens (Dashboard, Analyze, History, Profile, Docs)
│   │   └── styles/         # Global styles and design system variables
│   └── package.json        # Frontend dependencies
├── backend/                # FastAPI Application Server
│   ├── app/
│   │   ├── routes/         # HTTP endpoints (audio upload, predictions, contact, health)
│   │   ├── services/       # Audio conversion (FFmpeg), ML inference, and DB services
│   │   ├── utils/          # Config parser, logging, and JWT verifier (HS256/ES256/RS256)
│   │   └── main.py         # FastAPI App Entrypoint
│   └── requirements.txt    # Python dependencies
└── ml_pipeline/            # TensorFlow Model & Data Pipeline
    ├── saved_models/       # Pretrained CrySense CNN model (.keras)
    └── training/           # Model training and evaluation scripts

---

🛠️ Getting Started

Prerequisites

• Node.js (v18 or higher)
• Python (v3.11 or higher)
• FFmpeg (The backend uses static-ffmpeg to automate paths, but local installation is recommended for fallback)

---

Setup Instructions

1. Setup your Database (Supabase)

Create a new project on Supabase and run the supabase_migration.sql file in the SQL Editor to set up the schema:

• profiles: Stores user names and timestamps linked to auth.
• predictions: Stores classification results, timestamps, and confidence scores.

Enable Google Provider under Authentication -> Providers -> Google in the Supabase Dashboard if you wish to support Google OAuth login.

2. Configure Environment Variables

Create a .env file in the root folder:

CRYSENSE_ENVIRONMENT=development
CRYSENSE_SUPABASE_URL=https://<your-project-ref>.supabase.co
CRYSENSE_SUPABASE_SERVICE_KEY=<your-service-role-key>
CRYSENSE_SUPABASE_JWT_SECRET=<your-jwt-secret-or-key-id>
CRYSENSE_CORS_ORIGINS=["http://127.0.0.1:5173","http://localhost:5173"]

Create a .env file in the frontend/ folder:

VITE_SUPABASE_URL=https://<your-project-ref>.supabase.co
VITE_SUPABASE_ANON_KEY=<your-anon-key>
VITE_API_BASE_URL=http://127.0.0.1:8000

Note: The backend verifier automatically handles both legacy symmetric HS256 and modern asymmetric ES256 / RS256 keys generated by newer Supabase projects.

---

Running the Project

Start the Backend API

1. Navigate to the root directory and activate your Python virtual environment:

   # Windows
   .\.venv\Scripts\activate

2. Install dependencies:

   pip install -r backend/requirements.txt

3. Run the Uvicorn development server:

   python -m uvicorn backend.app.main:app --reload

   The backend API will be running at http://127.0.0.1:8000. You can inspect documentation at http://127.0.0.1:8000/docs.

Start the Frontend App

1. Navigate to the frontend/ directory:

   cd frontend

2. Install dependencies:

   npm install

3. Run the Vite development server:

   npm run dev

   Open http://127.0.0.1:5173 in your browser to view the application.

---

🔒 Security & Verification

• Token Verification: API endpoints are secured using Bearer JWT authentication linked to Supabase users.
• Security Headers: Integrated FastAPI middleware enforces security policies (XSS blocks, Frame restrictions, MIME sniffing protection, CORS limits).
• Build Optimization: The frontend utilizes Vite's Rollup configuration to compile assets cleanly, yielding a lightweight CSS size (~34.5 KB) and optimal bundle sizes.

---

📄 License

This project is private and proprietary. All rights reserved.