As one of the world’s largest and most multilingual transportation system, Indian Railways depends upon public address systems for the seamless functioning and safety of passengers. The Indian Railways suffers from poor voice quality, lack of personalization, limited contemporaneous adaptability, and insufficient support for multilingual functionalities. This is why we developed RailVani, a Generative AI-based public announcement system for Indian Railways. RailVani operates on a modular pipeline that includes data extraction from railway databases, inconsistency normalization, context enhancement using LLMs, and real-time speech generation with best of class TTS models. Additional features include multilingual outputs and customizable automation templates specific to individual stations. Leveraging the NISQA framework to evaluate speech discernment and intelligibility, a comparative analysis of four TTS models: Zonos, GPT-4o Mini, MMS-TTS, and Kokoro was assessed. The Kokoro and GPT-4o models showed unparalleled resilience in delivering value across varying acoustic conditions for advanced multilingual AI speech while traditional recordings failed obliterated the competition in all other metrics, proving the AI models’ spectacular battleground supremacy. The results confirm that RailVani improves the quality of announcements, responsiveness, and overall commuter satisfaction. Further research will include edge computing, TTS models resilient to noise interference, and adaptation to regional languages to broaden the system’s capabilities for public communication at Indian railway stations.