Language barriers continue to hinder seamless real-time digital communication, as most existing messaging platforms depend on external or context-agnostic translation mechanisms that disrupt conversational continuity and reduce semantic accuracy. This paper presents GloChat, a real-time multilingual chat system that integrates context-aware, AI-driven translation directly into the messaging pipeline. The system combines WebSocket-based communication with asynchronous translation powered by large language models, enabling immediate message delivery while translations are generated in a non-blocking manner. Unlike conventional sentence-level approaches, GloChat incorporates recent conversational context into the translation process, producing more accurate and less ambiguous outputs that better preserve intent, tone, and informal language patterns. Each participant interacts in their preferred language, with translations dynamically adapted for the recipient. The architecture is designed to support low-latency communication and scalable real-time interaction through decoupled processing and efficient message handling. Observations from the implemented system indicate that context-aware translation improves conversational coherence and user experience, demonstrating the practicality of integrating conversation-sensitive translation into modern messaging systems.
Abhishek Bharti (Thu,) studied this question.