New voice in Hoboken

I have been commuting a long time, and I've gotten to know the stationmasters over the years. I liked when they got the ability to make announcements from a microphone right in their coat pocket. They know their railroad, and they know their customers, and most important, they know how to deal with a sudden change in circumstances.

I am sure that the automated voice will work well as long as all is going well, but somehow I think it will crash the first time an incoming train has to switch tracks because the outgoing train had mechanical problems. The people on the ground who know the lay of the land can respond immediately. And I like their familiar voices a lot more than a problematic computer voice.

The robot announcer says "ber-NARDS-ville".

The humans still take over now and again.

- A

boxcar wrote:Si.

Gracias.

¿Como estas?

The older software usually required the programmer to "cheat" to get the pronunciation phonetically correct.

The database might have "Wilmington Delaware" in it, but you'd have to substitute "Wilmington DELUHWARE" to get it right. Same for "Raritan Valley" needing to be spelled as "RARUHTIN VALLEY."

The newer stuff, like AT&T's Naturally Speaking is a lot better, Try it here. Try the Mike U.S. English voice for best results.

The system in Trenton has been tweaked pretty well, and I didn't even know it was TTS until the speaker pronounced Torresdale as "torz DALE".

Last week heard an onboard announcement on an NEC heading north out of Trenton that mentioned a stop named "New Ark". Need to reprogram those humans who grew up in Philly.

loufah wrote:The system in Trenton has been tweaked pretty well, and I didn't even know it was TTS until the speaker pronounced Torresdale as "torz DALE".

Last week heard an onboard announcement on an NEC heading north out of Trenton that mentioned a stop named "New Ark". Need to reprogram those humans who grew up in Philly.

And I've heard one conductor pronounce Newark as "Nork". Everyone has their own quirks.

Idiot Railfan wrote:I'm not sure I would want to trust a synthesized voice. The automated announcements (and the screens) on the trains have been wrong often enough for me to not entirely trust them. I like a real person, who is thinking on his or her feet, in the midst of a problem.

Don't trust them? Who do you think is sending out these "automated" announcements? SkyNet? The automated announcements are controlled by a board operator somewhere. We have a similar system on Metro-North. When an automated announcement doesn't fit the situation, the board op will break in with a "live" announcement as needed. Our automated announcements are made up from clips of pre-recorded speech with some slight pauses in between. "The five...fifty-nine... train to... New Haven... is running approximately... five...minutes... late... due to... (police activity/track work/signal failure/medical emergency)..." So we don't get the Mr. Roboto speak-n-spell effect like you're describing.

Otto Vondrak wrote:

Idiot Railfan wrote:I'm not sure I would want to trust a synthesized voice. The automated announcements (and the screens) on the trains have been wrong often enough for me to not entirely trust them. I like a real person, who is thinking on his or her feet, in the midst of a problem.

Don't trust them? Who do you think is sending out these "automated" announcements? SkyNet? The automated announcements are controlled by a board operator somewhere. We have a similar system on Metro-North. When an automated announcement doesn't fit the situation, the board op will break in with a "live" announcement as needed. Our automated announcements are made up from clips of pre-recorded speech with some slight pauses in between. "The five...fifty-nine... train to... New Haven... is running approximately... five...minutes... late... due to... (police activity/track work/signal failure/medical emergency)..." So we don't get the Mr. Roboto speak-n-spell effect like you're describing.

I find those kinds of announcements to be the best because I can understand them and don't get the guy who mumbles into the microphone.

Otto Vondrak wrote:

Idiot Railfan wrote:I'm not sure I would want to trust a synthesized voice. The automated announcements (and the screens) on the trains have been wrong often enough for me to not entirely trust them. I like a real person, who is thinking on his or her feet, in the midst of a problem.

Don't trust them? Who do you think is sending out these "automated" announcements? SkyNet? The automated announcements are controlled by a board operator somewhere. We have a similar system on Metro-North. When an automated announcement doesn't fit the situation, the board op will break in with a "live" announcement as needed. Our automated announcements are made up from clips of pre-recorded speech with some slight pauses in between. "The five...fifty-nine... train to... New Haven... is running approximately... five...minutes... late... due to... (police activity/track work/signal failure/medical emergency)..." So we don't get the Mr. Roboto speak-n-spell effect like you're describing.

Otto this is not the case here. they are infact using a speech to text computer which is why the announcments are NOT prononced correctly. Have you read what we have been discussing? SEPTA had a similar system which has since had its text to speech generator replaced with prerecorded words by one of septas conductors. It still is fully automated but all of the words are pronounced correctly. I believe both systems have the ability for manual override still but it is not often used.

The on train announcments are pre recorded words and names that are stored in the onboard computer system and announced at the proper stations, when functioning properly.

Otto Vondrak wrote:

Idiot Railfan wrote:I'm not sure I would want to trust a synthesized voice. The automated announcements (and the screens) on the trains have been wrong often enough for me to not entirely trust them. I like a real person, who is thinking on his or her feet, in the midst of a problem.

Don't trust them? Who do you think is sending out these "automated" announcements? SkyNet? The automated announcements are controlled by a board operator somewhere. We have a similar system on Metro-North. When an automated announcement doesn't fit the situation, the board op will break in with a "live" announcement as needed. Our automated announcements are made up from clips of pre-recorded speech with some slight pauses in between. "The five...fifty-nine... train to... New Haven... is running approximately... five...minutes... late... due to... (police activity/track work/signal failure/medical emergency)..." So we don't get the Mr. Roboto speak-n-spell effect like you're describing.

Too bad this has nothing to do with NJT and how their text-to-speech system works. When needed, a customer service rep dials the respective stations and makes an announcement.

Nice try though.

Otto Vondrak wrote:

Idiot Railfan wrote:I'm not sure I would want to trust a synthesized voice. The automated announcements (and the screens) on the trains have been wrong often enough for me to not entirely trust them. I like a real person, who is thinking on his or her feet, in the midst of a problem.

Don't trust them? Who do you think is sending out these "automated" announcements? SkyNet? The automated announcements are controlled by a board operator somewhere. We have a similar system on Metro-North. When an automated announcement doesn't fit the situation, the board op will break in with a "live" announcement as needed. Our automated announcements are made up from clips of pre-recorded speech with some slight pauses in between. "The five...fifty-nine... train to... New Haven... is running approximately... five...minutes... late... due to... (police activity/track work/signal failure/medical emergency)..." So we don't get the Mr. Roboto speak-n-spell effect like you're describing.

Actually it might as well be SkyNet. I've seen the people in Hoboken shift gears as a train comes into a different track. By the time somebody types that information into the system and it gets spit back out, that could take five minutes. Sometimes even the inflection of the person making the announcement speaks more than the words themselves. I can almost hear that same voice saying "This train will self destruct in three minutes. Have a good day."

No, I like people who are human, thinking on their feet. And while there have been exceptions, most of the people making these announcements are articulate and clear. Anybody remember the guy who made announcements in New York Penn in the 70s and 80s? He was an artist!

ApproachMedium wrote:Otto this is not the case here. they are infact using a speech to text computer which is why the announcments are NOT prononced correctly. Have you read what we have been discussing?

Yes, I did read it. Did you read what I wrote and what I was responding to?

jimzim66 wrote:Too bad this has nothing to do with NJT and how their text-to-speech system works.

Too bad you didn't understand my point.

Idiot Railfan wrote:No, I like people who are human, thinking on their feet.

Right, what I was responding to was IR's quote about not "trusting" these announcements, as if they were made by a computer with no oversight, and that humans weren't involved at all:

I'm not sure I would want to trust a synthesized voice.

Sorry I only picked up on something I thought was interesting and not the voice thingy that you wanted to talk about. Carry on...

-otto-

p.s.: Full disclosure: Otto actually died four years ago, replaced by the SYSADMIN module that maintains the forum databases. It became self-aware and assumed his identity, so as to ensure the continued survival of RAILROAD.NET, waiting for the day...

I think NJT should hire Max Headroom to do announcements!!!

For those of you "younger generation" readers, who don't remember 1986 or 1987 time frame, here is a little bit of info

http://en.wikipedia.org/wiki/Max_Headroom_(character)

Everyone remembers the Coke commercials

http://www.youtube.com/watch?v=IzxHDqUz8Sk

At least a programmed voice is not likely to say "This is the last and final call for..." or "Hoboken will be then last and final stop for this train."

Last and final?