This document outlines a perceptually adaptive audio system designed to improve the intelligibility of spoken announcements in highly reverberant environments. Using speech buffering, pacing modification, spectral shaping, and real-time feedback, the system enhances clarity without architectural intervention. The system is privacy-respecting, low-cost, and requires no retrofitting. It is also applicable to real-time interpersonal communication in noisy or complex environments. This is a timestamped public disclosure under a Creative Commons licence.