Terminal C's main concourse was 300 meters long with ceiling speakers every 10 meters. Each group of speakers was supposed to be individually delayed so that a gate announcement would travel down the corridor in sync with the speed of sound, creating a seamless wave of intelligible speech. The installer programmed all 30 zones with the same delay. A different delay for each zone, but each one was wrong by a different amount.

Walking down the concourse, you could hear the same announcement start, stop, restart, overlap itself, and restart again — like walking through layers of time. Zone 1 was 30ms early. Zone 5 was on time. Zone 12 was 80ms late. Zone 20 was 45ms early again because the installer had accidentally looped the delay values. A single 10-second announcement took 25 seconds to fully propagate and created, at any given point, three overlapping versions of itself.

Passengers described it as "walking through an announcement that was having an argument with itself." Flight information was extracted through a combination of lip-reading the gate agents, checking the departure board, and triangulating which of the three overlapping voices seemed most confident.

Long corridors with distributed speakers require delay values calculated from the speed of sound (343 m/s) and the distance between zones. Every 10 meters requires approximately 29ms of additional delay.

The Moral: Distributed speaker systems require precise delay calculation for every zone. Use SonaVyx Transfer Function impulse response at each zone boundary to verify timing — wrong by even 20ms creates audible echo.