Talky - Real-time Chat Application

Features

• Real-time messaging with WebSocket connections
• Public and private chat functionality
• AI-powered chat summarization using BART model
• User management with online status
• Responsive design for desktop and mobile

Live Demo

Demo: https://talky-chat-eight.vercel.app/

Note: The summarization feature is disabled on the live demo due to memory limitations. Run locally for full functionality.

Tech Stack

• Frontend: React, WebSockets, CSS3
• Backend: Node.js, WebSocket Server
• AI Service: FastAPI, Transformers, BART model

Local Setup

Installation

1. Clone repository

git clone https://github.com/sub-aina/live-chat.git
cd talky-chat

2. Frontend setup

cd client
npm install

3. Server setup

cd server
npm install uuid ws node-fetch

4. AI service setup

cd summarize
pip install fastapi uvicorn transformers torch pydantic

Running the Application

Run these commands in separate terminals:

1. Start AI service

cd summarize
python summarizer_api.py

2. Start WebSocket server

cd server
node index.js

3. Start frontend

cd client
npm run dev

Configuration

Environment Variables

• PORT: Server port (default: 8001 for WebSocket, 8000 for API)
• SUMMARIZER_API_URL: Summarization API endpoint

Frontend Configuration

Update WebSocket URL in client/src/components/Home.jsx:

const WS_URL = `ws://127.0.0.1:8001`;