A Survey on Speech Enhancement Techniques for Improved Translation in Multilingual Acoustic Environments
Geethika Pula, Manikumar V S S S R, D. Sasikala · 2025
Speech-to-Speech Translation (S2ST) systems are inevitable tools in real-time communication to break language barriers. The design and development of an S2ST system is specifically for public meetings, where participation in the public meeting is typically multilingual. This paper discusses the latest developments in the S2ST system, specifically language translation and speech recognition, with a focus on various accents, dialectal variations, and contextual nuances, making translation accurate and natural. The work includes how domain-specific datasets improve quality in translation, neural models that can process more efficiently, and aims to make people accessible, provide access, and encourage multilingual collaboration in public forums. Future work will focus on expanding language support, continuing to lower latency even further, and introducing emotion-aware synthesis that is increasingly meant to capture an audience.