tom1649
Member
- Joined
- 5 Jul 2010
- Messages
- 989
At Lincoln, a bell used to ring when a train was approaching. If there were any delays the ticket office staff would do the honours.
-
Our new ticketing site is now live! Using either this or the original site (both powered by TrainSplit) helps support the running of the forum with every ticket purchase! Find out more and ask any questions/give us feedback in this thread!
Automated Announcements
- Thread starter thefab444
- Start date
- Status
At Lincoln, a bell used to ring when a train was approaching. If there were any delays the ticket office staff would do the honours.
Sponsor Post - registered members do not see these adverts; click here to register, or click here to log in
R
RailUK Forums
Blindtraveler
Established Member
- Joined
- 28 Feb 2011
- Messages
- 10,544
bad as this text to speach system prob is I
have to agree with John morris that ann is just, well, not my thing. The acception has to be LM's system which is at least accurate and works
have to agree with John morris that ann is just, well, not my thing. The acception has to be LM's system which is at least accurate and works
Philip
On Moderation
Is Chester the most northerly station that Ruth (ATW woman) can be heard at or are there any further north?
Also, do they use the woman with the Yorkshire accent at all on 185s now? She used to do safety stuff, CCTV/security/no smoking stuff and also used to remind customers to have tickets ready for the barrier at Oxford Road - the main stopping pattern announcer now does the exact same announcement.
Speaking of the usual 185 announcer, I used to like how she just announced the train destination and the next stop when a train left a station, nice and quick - nowadays she goes through all the stations after every stop...
Also, do they use the woman with the Yorkshire accent at all on 185s now? She used to do safety stuff, CCTV/security/no smoking stuff and also used to remind customers to have tickets ready for the barrier at Oxford Road - the main stopping pattern announcer now does the exact same announcement.
Speaking of the usual 185 announcer, I used to like how she just announced the train destination and the next stop when a train left a station, nice and quick - nowadays she goes through all the stations after every stop...
I think text to speech is the future, and it's very nearly there. The system used by LUL at King's Cross is okay, but they've got to continue to tweak it, especially when it does long sentences and has the pauses in the wrong place.
It makes much more sense to work to this as a solution as it means you can easily change announcements, or customise them, to suit your needs and not need (sorry Phil etc) to hire a voice artist and re-hire for updates.
The key to successfully pronouncing station names (or even TOC names) is to have a look-up table that will give the system the phonetic data for a name. Therefore it will always be right, something that isn't always the case on a SatNav reading place or road names. Again, this is now happening with modern maps containing this data, if the satnav manufacturer chooses to pay for it.
There are quite a lot of online TTS (text-to-speech) engines around, but here's one that has lots of voices. Unfortunately you can't enter a whole announcement as there's a character limit, but it's a bit of fun (try Damian or Vixen, or ransom to hear mixed voices)
I had mine saying: The train approaching platform one does not stop here. Please stand back from the platform edge.
http://www.voiceforge.com/demo
Incidentally, TTS also makes it possible to do multi-lingual announcements on an ad-hoc basis if necessary.
And another site:
http://text-to-speech.imtranslator.net/
http://tts.imtranslator.net/GqfW - example announcement from Hatfield
http://tts.imtranslator.net/Gqfm - made up running and engineering information (misspelling of WGC is deliberate!)
I can't see how any TOC would (or should) be investing in a system that requires updated voice recordings, editing etc. I am sure Phil Sayer could provide the voice data for one of the TTS companies too, if he hasn't already.
It makes much more sense to work to this as a solution as it means you can easily change announcements, or customise them, to suit your needs and not need (sorry Phil etc) to hire a voice artist and re-hire for updates.
The key to successfully pronouncing station names (or even TOC names) is to have a look-up table that will give the system the phonetic data for a name. Therefore it will always be right, something that isn't always the case on a SatNav reading place or road names. Again, this is now happening with modern maps containing this data, if the satnav manufacturer chooses to pay for it.
There are quite a lot of online TTS (text-to-speech) engines around, but here's one that has lots of voices. Unfortunately you can't enter a whole announcement as there's a character limit, but it's a bit of fun (try Damian or Vixen, or ransom to hear mixed voices)
I had mine saying: The train approaching platform one does not stop here. Please stand back from the platform edge.
http://www.voiceforge.com/demo
Incidentally, TTS also makes it possible to do multi-lingual announcements on an ad-hoc basis if necessary.
And another site:
http://text-to-speech.imtranslator.net/
http://tts.imtranslator.net/GqfW - example announcement from Hatfield
http://tts.imtranslator.net/Gqfm - made up running and engineering information (misspelling of WGC is deliberate!)
I can't see how any TOC would (or should) be investing in a system that requires updated voice recordings, editing etc. I am sure Phil Sayer could provide the voice data for one of the TTS companies too, if he hasn't already.
Last edited:
thefab444
Established Member
The down side is that widespread use of TTS technology will almost certain to kill off the automated announcements community, although some might say this is a good thing. Certainly, I would lose interest and I know a few others would as well.
I'm not sure that's a good reason to stop progress, but there would still be scope to listen to different voices. TTS doesn't mean a single monotone voice anymore.
Given the freedom to have systems say anything, you could even argue that there would be more reason to travel around to see different configurations of the system, rather than the likelihood that most TOCs will end up with the Atos Origin system with a single voice and standard phrases.
Given the freedom to have systems say anything, you could even argue that there would be more reason to travel around to see different configurations of the system, rather than the likelihood that most TOCs will end up with the Atos Origin system with a single voice and standard phrases.
I heard ATOS Anne in use at Nottingham last week - she was telling passengers to take care due to the inclement weather (it was sunny!) & seemed to be going on almost non-stop.
A big difference to the very clear manual announcements being made at Grantham.
A big difference to the very clear manual announcements being made at Grantham.
Speaking of Nottingham, has anyone heard some other manually activated announcements there? When I was waiting there one evening a few weeks ago, a manual announcement was made asking a passenger to return to their luggage. This was followed by a set of chimes, then a recorded security announcement was played, not the usual ATOS Anne one.
I've heard similar announcements in use when Anne was (presumably) broken, and the station staff were doing manual announcements. I'm sure I've also heard this system remind passengers not to use train toilets in the station... though I may have been dreaming that!
I've heard similar announcements in use when Anne was (presumably) broken, and the station staff were doing manual announcements. I'm sure I've also heard this system remind passengers not to use train toilets in the station... though I may have been dreaming that!
When FCC got the system, it was going on and on and on. It took months to finally get things sorted, and when I asked on a recent FCC forum why Hatfield now has all platform announcements on all speakers (most annoying to be in the station building being told to stand clear of platforms on the other side of a bridge), I was told it is too expensive to configure it for different zones at the smaller stations.
Now given it was zonal before and it's just software at the end of the day, why the hell does FCC have to pay Atos more money to configure it the same way it used to be? It's not as if the speakers are all connected to one audio output (well, they weren't before!)
WestCoast
Established Member
Text To Speech is far more advanced in the airport announcement market, with a company called Aviavox leading the way.
www.aviavox.com/GB - samples available (bottom right hand corner).
Quality artificial voices (in numerous languages) with artificial intelligence. Has been installed at a number of UK airports.
I could see this company's products being capable of decent rail announcements.
The problem is the Atos system seems to be cheap and presumably recommended by Network Rail, which in turn means a TOC probably has little incentive to use anything else.
Indeed, I wonder if that (given stations are owned by Network Rail) they're effectively being forced to buy the Atos system.
Given the size of Atos, I am surprised they can't build a system that isn't much better, or even introduce a TTS system of their own.
Indeed, I wonder if that (given stations are owned by Network Rail) they're effectively being forced to buy the Atos system.
Given the size of Atos, I am surprised they can't build a system that isn't much better, or even introduce a TTS system of their own.
thefab444
Established Member
Whilst I'm sure TTS could do the job well, I've yet to hear a system that sounds quite like a real person!
As for that Aviavox system, it sounds a bit poor to me because it's just replacing segmented human speech with segmented computerised speech! If it has to be TTS, it has to be Acapela's Peter.
As for that Aviavox system, it sounds a bit poor to me because it's just replacing segmented human speech with segmented computerised speech! If it has to be TTS, it has to be Acapela's Peter.
Does it have to sound like a perfect human voice? It just needs to be clear. I am not sure TTS is trying to become 100% human like, even to the point where it might alter the speech and add pauses/stutters/mistakes to fool people.
From an old computer speech synthesis program on a Commodore 64 many years ago, there are ways to change the pitch and tone and emphasis certain words. Obviously for the online demo engines I've found, this isn't possible (or I don't know what they are) but this all helps make them more natural.
As the systems improve, this will become less necessary. That and extra data for each place name to ensure the correct pronunciation.
From an old computer speech synthesis program on a Commodore 64 many years ago, there are ways to change the pitch and tone and emphasis certain words. Obviously for the online demo engines I've found, this isn't possible (or I don't know what they are) but this all helps make them more natural.
As the systems improve, this will become less necessary. That and extra data for each place name to ensure the correct pronunciation.
thefab444
Established Member
Well, people complain about automated announcements using human speech segments, goodness knows what they'd think of TTS usage en masse.
I fail to see how this isn't a good thing.
Say I was in the ticket office and was on my way to that platform, how would I know to standa clear if I didn't hear the announcement? Announcements should be made on all platforms in my opinion as if someone is on the wrong platform, they may have a chance to get to the right one as the train is pulling in. I know some stations are too busy for this, but that is a different case all together!
So you're buying your ticket and need to be told to stand away from the edge of platform 2 (or 3). Given platform 2 is the down fast, do you really need to be told over and over? It must seriously annoy gateline staff and anyone else waiting would eventually switch off, especially when they use the male voice to tell you how to be careful when getting on or off a train that isn't there.
Yes, you do want to be told about trains approaching, except FCC have stopped doing advance announcements so by the time you hear the one and ONLY announcement for your train going north, you've got around 20-30 seconds to get there. Given the trains stop at the far end on platform 3, good luck - or be damn careful if you choose to run!
If you waited for that announcement then, sorry, you've missed the train.
Only running information should be broadcast in the ticket hall (as it USED to be).
Oh, and there aren't currently any working speakers on platforms 2/3 - bar a low volume one that squeaks near the stairs. It's been like that since September. It's now almost August. FCC has been told multiple times about this and they've done nothing. So, perhaps you should wait on the other side after all and then run!
Yes, you do want to be told about trains approaching, except FCC have stopped doing advance announcements so by the time you hear the one and ONLY announcement for your train going north, you've got around 20-30 seconds to get there. Given the trains stop at the far end on platform 3, good luck - or be damn careful if you choose to run!
If you waited for that announcement then, sorry, you've missed the train.
Only running information should be broadcast in the ticket hall (as it USED to be).
Oh, and there aren't currently any working speakers on platforms 2/3 - bar a low volume one that squeaks near the stairs. It's been like that since September. It's now almost August. FCC has been told multiple times about this and they've done nothing. So, perhaps you should wait on the other side after all and then run!
- Status