Paco (2026: New) Hope@paco@infosec.exchange
Took the #train to work today, which I do when I go to Arlington. First time I’ve seen #VRE get so screwed up. 2 trains due a while ago are 40+ minutes late.
The voice announcer is automated text to speech. But it’s like Dr Sbaitso bad. We have had far better text-to-speech for decades. It has silly quirks that show it has not been tuned for public transport.
Train 305 is referred to as “train three hundred and five,” which is not how anyone would say that.
It consistently emPHAsizes the wrong sylLABles.
With text-to-speech system in the age of very accurate machine translation, we could be making these announcements in several languages. But no. We only do English and with a cadence that even native speakers struggle to understand.
This was state of the art in 1992. Compare this to what you hear on a modern train or airport text to speech.
https://m.youtube.com/watch?v=sV3pYZZ2jEw
Compare with the current VRE:
https://m.youtube.com/shorts/khSbA09stEg. Especially how it says “VRE” at the end of “thank you for riding the VRE”
Compare that to metro in DC which seems to have some modern speech that is quite smooth.
https://m.youtube.com/shorts/MCpx-ur7J-U
And finally, our national embarrassment, Washington Dulles International Airport. A lot of it is smooth. But when it comes to something essential like “Board here for all B gates,” you get “bored here for albigates.”
https://m.youtube.com/shorts/rCTWEs6wQI0