Translate, Now! Near Real-Time Speech-to-Speech Translation with Urgency Preservation

In an increasingly connected world, where people from different countries work closely together to pursue common goals, language barriers remain a significant obstacle to seamless communication. This issue becomes even more critical in emergency situations, where urgent messages must be conveyed immediately to multiple recipients, minimizing the risk of misunderstanding due to limited knowledge of the local language. To address this challenge, we propose a novel system for near real-time voice translation that preserves the tone of urgency. Our system incorporates advanced artificial intelligence components for speech recognition, urgency detection, machine translation, and speech synthesis. Being specifically tailored for urgency detection and urgency-aware translation, it enhances state-of-the-art translation technologies by preserving the urgency tone, which is often lost in conventional systems. We developed a working prototype of our system and conducted experiments that show the efficiency and low latency of our solution.