https://github.com/andreasholt/multilingual-chatroom-app

Realtime (WebSockets-based) Crossplatform "Multilingual" Chatroom Application
https://github.com/andreasholt/multilingual-chatroom-app

Last synced: about 1 month ago
JSON representation

Realtime (WebSockets-based) Crossplatform "Multilingual" Chatroom Application

• Host: GitHub
• URL: https://github.com/andreasholt/multilingual-chatroom-app
• Owner: AndreasHolt
• Created: 2024-08-11T16:06:09.000Z (6 months ago)
• Default Branch: master
• Last Pushed: 2025-01-01T19:55:16.000Z (about 1 month ago)
• Last Synced: 2025-01-01T20:31:34.240Z (about 1 month ago)
• Language: C#
• Homepage:
• Size: 3.24 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Real-time "Multilingual" Chatroom App

**Motivation**: While traditional translation APIs perform word-for-word translations, they often miss the context and nuances of casual conversations, struggling with slang, sayings, and everyday expressions. This project explores using LLMs for more natural and context-aware translations, enabling people to chat naturally across language barriers in real-time.

## Demo

![Demo](demo.gif)

**Intended usage**: User either joins a room with a room ID, or creates a new room. The user selects their language of preference prior to joining. When a user sends a message in their selected language, it is received by all other users in the room, in their own preferred language - allowing them to communicate through language barriers.

## Features

- A chatroom application featuring real-time messaging (WebSocket-based) with language translation using LLMs (fast inference via Groq's API)

  - Use LLM's such as Llama 3.1 8b, Mixtral 8x7b and Gemma 7b

- Crossplatform: Windows, OSX, Linux and Web (WebAssembly)

- Utilizes Avalonia UI for the frontend, and .NET Core for the backend

- Retro Windows 9.x UI because it's fun

## LLMs for Translation vs Traditional Translation APIs

Pros:

- While translation APIs can provide direct word-for-word translations, they may miss contextual meaning and informal language that's common in chat messages. For example, slang and abbreviations

- Better handles conversational flow and tone

Cons:

- Slower than dedicated translation APIs (especially for larger models) and higher computational cost

- May hallucinate