VoxBridge: Bridging Voices, Breaking Language Barriers

The increasing demand for global communication has highlighted the limitations of existing real-time communica- tion systems, particularly in handling multilingual interactions. Although technologies such as WebRTC enable efficient low- latency communication, they lack integrated translation capa- bilities. This paper presents VoxBridge, a real-time audio broadcasting system that integrates speech recognition, machine translation, and speech synthesis into a unified pipeline. The system leverages WebRTC and mediasoup for audio transmission, Whisper for speech recognition, NLLB-200 for translation, and Edge-TTS for speech synthesis. The architecture is designed using Node.js, Express, Socket.IO, and gRPC to ensure modularity, scalability, and low latency. Experimental analysis demonstrates that the system achieves near real-time performance with acceptable latency and high accuracy, making it suitable for multilingual communication scenarios such as online education, global meetings, and broad- casting systems.