Get started

Run the project locally: clone the repo, set up the backend and frontend, configure optional environment variables, and deploy to production when ready.

API referenceDetect

Overview

The project has two parts: a backend (Flask + PyTorch + RawNetLite) that loads the model and serves /health and /predict, and a frontend (Next.js) that proxies those to /api/health and /api/predict. You run both locally for development.

Prerequisites

  • Python 3.8+ (for backend)
  • Node.js 18+ and npm (for frontend)
  • Git (to clone the repository)
  • Model checkpoint (.pt) and RawNetLite.py in the backend

Installation

Clone the repository and install dependencies for backend and frontend.

git clone https://github.com/zuberkhan01st/Vcs-DeepFake-Audio-Detection
cd Vcs-DeepFake-Audio-Detection

Backend

Create a virtual environment, install dependencies, and place the model file in backend/models/.

cd backend
python -m venv venv
# Windows: venv\Scripts\activate
# macOS/Linux: source venv/bin/activate
pip install -r requirements.txt
# Place your .pt file in backend/models/
# Ensure RawNetLite.py is in backend/

Frontend

From the project root, install and run the Next.js app.

cd web
npm install

Configuration

Optional environment variables:

  • Backend: MODEL_ROOT (default: ./models), BEST_MODEL (default: augmented_triple_cross_domain_focal_rawnet_lite.pt), PORT (default: 5000).
  • Frontend: BACKEND_URL in .env.local (e.g. http://127.0.0.1:5000) if the backend is not on the default.

Download & model

The model checkpoint (.pt) must be in backend/models/ (or MODEL_ROOT). Default filename: augmented_triple_cross_domain_focal_rawnet_lite.pt. You also need RawNetLite.py (model definition) in the backend. Check the project README or release assets for download links.

Running the app

Start the backend first, then the frontend.

Terminal 1 (backend)

cd backend
# with venv activated
python app.py

Terminal 2 (frontend)

cd web
npm run dev

Open http://localhost:3000. Use Detect to upload audio and get predictions.

Production deployment

Run Flask behind a WSGI server (e.g. Gunicorn) and the Next.js app behind a reverse proxy (e.g. Nginx). Use environment variables for BACKEND_URL, MODEL_ROOT, and secrets. Docker support may be added in a future release.

Next steps

  • Deployment — host locally or on a VPS, Nginx, SSL, security.
  • API reference — endpoints, schemas, code examples.
  • Mission — why we build this and how to contribute.