The Problem

I was waiting for BART today when there was a station announcement made over the PA system. Between the noise on the platform and the poor quality of the speakers, it was very difficult to understand the content of the announcement - this is a pretty common occurrence in any public forum where announcements are made over speakers.

The Solution

I wonder if there's an optimal accent that could be used to convey information in these high noise environments. Perhaps an ideal accent would articulate / exaggerate all the phonemes (no dropped 'r's, the 'h' following a 'w') and exacerbate the difference between similar sounding phonemes. This would help listeners disambiguate between potentially similar words.

Another possible strategy is to use more multi-syllable words. Longer words are less likely to be mistaken for other words and people are can still still be understood if one or two syllables are drowned out by noise.

Accents and speech patterns optimized in this way are likely to sound pretty odd, but they could immensely increase understanding. Announcers could be trained to speak this way but it's probably easier to use text to speech programs.

Bonus Idea

(via Chris Maury)

If the noise tends to be around a common frequency (low rumbles in a train station, human voices in a crowded area), then one could modulate the announcements to give them a quieter part of the spectrum. This should allow people to more easily isolate the announcements.