Sad Sprinter
Established Member
Which one of these files contain Cell by the way?
-
Our booking engine at tickets.railforums.co.uk (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!
Ketech Cellia Drummond and Only Woman announcements publicly availible?
- Thread starter polskism
- Start date
Sponsor Post - registered members do not see these adverts; click here to register, or click here to log in
R
RailUK Forums
It is the "female1" folder that contains all of the Celia Drummond sound files.
These are which folders contain what:
"female1" folder - Celia Drummomd
"male1" folder - Phil Sayer
"female2" folder - Alison Mckay (Scotrail)
"male3" folder - Michael Comyn (Irish Rail)
"default" folder - Michael Comyn (Irish Rail)
They all contain thousands of files so it can take a while to find your required file.
MrJeeves
Established Member
I've transcribed up to file 8031 so far: https://github.com/Rail-Announcements/ketech-llpa-announcements/blob/main/id_to_script_mapping.csv
Stations are transcribed as their CRS where appropriate.
Stations are transcribed as their CRS where appropriate.
MrJeeves
Established Member
I'm happy to say that I've begun the work to implement Phil Sayer into my announcements website using the original recordings.
railannouncements.co.uk
At the moment, this is focused mainly on Amey/Ditra Southern style announcements, but I'm open to additional formats in the future.
Currently, only "Next train" announcements are available, but the logic and UI is in place for this to also handle multiple via points, short platforms, request stops, and also integrating these features into dividing services, too.
I've attached a couple of examples from the site.
Fictional West Coastway service with divide, request stops and short platforms:
View attachment short platform galore with split.mp3
CrossCountry to Brighton with multiple via points:
View attachment XC to BTN.mp3
Virgin Pendolino to Edinburgh with short platform:
View attachment VT to EDB.mp3
UK Rail Announcement Generator
Generate various station and on-train announcements for the UK rail network using raw audio recordings.
railannouncements.co.uk
At the moment, this is focused mainly on Amey/Ditra Southern style announcements, but I'm open to additional formats in the future.
Currently, only "Next train" announcements are available, but the logic and UI is in place for this to also handle multiple via points, short platforms, request stops, and also integrating these features into dividing services, too.
I've attached a couple of examples from the site.
Fictional West Coastway service with divide, request stops and short platforms:
View attachment short platform galore with split.mp3
CrossCountry to Brighton with multiple via points:
View attachment XC to BTN.mp3
Virgin Pendolino to Edinburgh with short platform:
View attachment VT to EDB.mp3
TT-ONR-NRN
Established Member
Please be careful, I’d hate it if Elinor Hamilton were to sue you for thatI'm happy to say that I've begun the work to implement Phil Sayer into my announcements website using the original recordings.
UK Rail Announcement Generator
Generate various station and on-train announcements for the UK rail network using raw audio recordings.
At the moment, this is focused mainly on Amey/Ditra Southern style announcements, but I'm open to additional formats in the future.
Currently, only "Next train" announcements are available, but the logic and UI is in place for this to also handle multiple via points, short platforms, request stops, and also integrating these features into dividing services, too.
I've attached a couple of examples from the site.
Fictional West Coastway service with divide, request stops and short platforms:
View attachment 145574
CrossCountry to Brighton with multiple via points:
View attachment 145575
Virgin Pendolino to Edinburgh with short platform:
View attachment 145576
MrJeeves
Established Member
Well, at least these files have been released into the public domain as a result of FOI!Please be careful, I’d hate it if Elinor Hamilton were to sue you for that
TT-ONR-NRN
Established Member
Ahh I see. Nice work!Well, at least these files have been released into the public domain as a result of FOI!
bramling
Veteran Member
Well, at least these files have been released into the public domain as a result of FOI!
The ones I’d really love to have are those of John Elgar, who was the predecessor to Phil Sayer especially in the Southern Region. I always thought he had the perfect voice, and IMO has never been bettered since in terms of sounding natural even when different segments have been stitched together. I know there are bits of his announcements out there, but I suspect we will never get a complete set of files, which is a shame.
Why does it say 'KeTech Phil Sayer' if its in the Amey style?I'm happy to say that I've begun the work to implement Phil Sayer into my announcements website using the original recordings.
UK Rail Announcement Generator
Generate various station and on-train announcements for the UK rail network using raw audio recordings.
At the moment, this is focused mainly on Amey/Ditra Southern style announcements, but I'm open to additional formats in the future.
Currently, only "Next train" announcements are available, but the logic and UI is in place for this to also handle multiple via points, short platforms, request stops, and also integrating these features into dividing services, too.
I've attached a couple of examples from the site.
Fictional West Coastway service with divide, request stops and short platforms:
View attachment 145574
CrossCountry to Brighton with multiple via points:
View attachment 145575
Virgin Pendolino to Edinburgh with short platform:
View attachment 145576
MrJeeves
Established Member
Because originally it wasn't but now it is, and I've not changed the name.
I think it's a slight hybrid of the two in reality, especially if you've only got one short platform (or request stop?).
I think it's a slight hybrid of the two in reality, especially if you've only got one short platform (or request stop?).
stan claire
Member
is there a Buckshaw and Chorley announcement file of Celia (and BSV for Anne)? wanting to use those but finding them is just about impossible
Strangely both Buckshaw Parkway and Chorley are missing from these sound files. I can not find them either. I am sure that Celia Drummond and Phil Sayer would have recorded these so it does seem odd. Perhaps they are newer recordings done after this batch of files was created.
Sebrh11
Member
DELETED
Last edited by a moderator:
DJ_K666
Member
Does anyone remember Bernie who used to do live announcements at Waterloo? She was in the 'Terminus' film in the 60s and then appeared in the 1983 one. She was a very well spoken lady indeed.
Sebrh11
Member
I've tried with no luck so I've asked someone with more skill. I'm yet to hear back
That's unfortunate, however it's nice to know people are actively trying to decode them. On another note are you aware if the Amey 'Only Woman' announcements are available anywhere? I'm aware it's been mentioned in the thread previously but not sure if anyone's found them
No unfortunately nobody has been able to. I have been trying for six years with no success. I am sure there is a way but it is just finding someone who knows how to access them. We originally were unable to access the .seg files for many months until someone on here found a solution so i presume there is a way to access the .pcc files too.
The only information we have is what TFL have told us about them:
The closest anyone has gotten to accessing the files is jonny5532 in post #81 in this thread. He managed to get some muffled sound out of it. But we have still been unable to properly work out how to access the files.
No sadly the Only Woman files have not been released yet. I hope we are able to get hold of them eventually as she was my favourite announcer. I have already spoken to Amey and they have had a good search and they no longer hold them. They disposed of all their sound files and equipment back in 2019 when they closed their CIS business. I was told by Amey that Ketech and TG Baker are both likely to still hold the files. Ketech own Ditra which is the old system that Amey ran off of so are quite likely to still hold the sound files. TG Baker are the company that Amey went through to get recordings. Amey never knew who Only Woman was because they went through TG Baker to get their recordings. So it is likely that TG Baker would also still hold the sound files. Unfortunately i have been unable to get in contact with Ketech or TG Baker and all of my emails to both of them get ignored. So unfortunately getting hold of the Only Woman sound files seems very difficult.
stan claire
Member
Buckshaw is a new(ish) station but Chorley has been open for ages. It couldn't be a new recording
Sebrh11
Member
I have asked an AI about the PCC files. It says"A PCC audio file refers to a file format that stands for "Pulse Code Modulated Audio File". PCC files contain audio recordings that have been digitally sampled and encoded using pulse code modulation. This format is commonly used for storing and transmitting high-quality audio data." So jonny5532 was right as it is a pcm formatted file.
2. Once you've found a suitable converter, visit their website and navigate to the PCC conversion" I am unsure if it will work but I will try that method.
The tricky part is converting the files to wav. The AI says for converting them "1. Open a web browser on your computer and search for an online PCC converter or file conversion software. Some popular options include Zamzar, FileZigZag, and Convertio.
2. Once you've found a suitable converter, visit their website and navigate to the PCC conversion" I am unsure if it will work but I will try that method.
MrJeeves
Established Member
Because AI is always 100% accurate and reliable...
...especially when such PCC files also contain data for the dot matrix displays, which are definitely pulse code modulated audio.
...especially when such PCC files also contain data for the dot matrix displays, which are definitely pulse code modulated audio.
Any
Any plans to make an Irish Rail announcement generator?
LdTransport203
New Member
Pretty sure the garbled announcments say 'This train will remain here a short time before departure' and '- should wait across platform at this station for a train to Terminal Fou-'
I know it's been a moment since anything has been added here, but out of interest, do you recall what settings you had when decoding this audio?
I'm currently at a similar stage where I can definitely hear the announcements amongst the muffled sound, but so far my experimentation hasn't led me anywhere further. I'm also trying to look into voice coding techniques but my audio experience is very limited for this sort of thing, so I'm not getting too far!
I didn't really get any further with that "copy file.pcc" format, here's a summary of what I learned:
The file seems to consist of chunks of audio data, each preceded by a 256-byte table (that may contain codec settings or playback metadata), all followed by some metadata chunks containing ASCII text (which may feed the in-carriage displays).
The audio data seems to be arranged in 16-bit words, since when viewed like that, bit (1<<13) is always 1. This is itself odd as it is a waste of bandwidth, and eliminates all of the audio codecs that I'm aware of, which don't do this.
I am able to extract extremely noisy but discernible audio by extracting bit (1<<6) from every 16-bit word and outputting as 1-bit PCM audio. It sounds much worse than correct 1-bit audio would do, but the roughly correct audio pitch suggests that the frame rate (at 16-bit frames) is around 4kHz.
16-bit frames at 4kHz means the bitrate is 64kbps, which is half that of the .seg files (which are uncompressed a-law PCM), and therefore implies some compression since the quality would be too bad otherwise.
The obvious candidate for a compression algorithm with a regular frame pattern and suitable bitrate is ADPCM, but it won't decode with any of the ADPCM codecs I have tried, and the stuck (1<<13) bit also doesn't make sense.
My guess is that it is either the binary input format for some off-the-shelf speech synthesis chip (although I've had a look around and haven't found any possible candidates), or it is a proprietary compression algorithm which isn't widely used elsewhere.
Here's a Python script to extract the noisy 1-bit audio:
Also attached is an mp3 of the whole file (as output by the above).
The file seems to consist of chunks of audio data, each preceded by a 256-byte table (that may contain codec settings or playback metadata), all followed by some metadata chunks containing ASCII text (which may feed the in-carriage displays).
The audio data seems to be arranged in 16-bit words, since when viewed like that, bit (1<<13) is always 1. This is itself odd as it is a waste of bandwidth, and eliminates all of the audio codecs that I'm aware of, which don't do this.
I am able to extract extremely noisy but discernible audio by extracting bit (1<<6) from every 16-bit word and outputting as 1-bit PCM audio. It sounds much worse than correct 1-bit audio would do, but the roughly correct audio pitch suggests that the frame rate (at 16-bit frames) is around 4kHz.
16-bit frames at 4kHz means the bitrate is 64kbps, which is half that of the .seg files (which are uncompressed a-law PCM), and therefore implies some compression since the quality would be too bad otherwise.
The obvious candidate for a compression algorithm with a regular frame pattern and suitable bitrate is ADPCM, but it won't decode with any of the ADPCM codecs I have tried, and the stuck (1<<13) bit also doesn't make sense.
My guess is that it is either the binary input format for some off-the-shelf speech synthesis chip (although I've had a look around and haven't found any possible candidates), or it is a proprietary compression algorithm which isn't widely used elsewhere.
Here's a Python script to extract the noisy 1-bit audio:
Code:
import wave
data = open("copy file.pcc", "rb").read()
out = wave.open(open("out.wav", "wb"))
out.setnchannels(1)
out.setsampwidth(1)
out.setframerate(4000)
out.writeframes(bytes([
(data[i]&0x40)<<1 for i in range(1, len(data), 2)
]))Also attached is an mp3 of the whole file (as output by the above).
Last edited:
The request stop that you implemented is just like the script that Atos use.I'm happy to say that I've begun the work to implement Phil Sayer into my announcements website using the original recordings.
UK Rail Announcement Generator
Generate various station and on-train announcements for the UK rail network using raw audio recordings.
At the moment, this is focused mainly on Amey/Ditra Southern style announcements, but I'm open to additional formats in the future.
Currently, only "Next train" announcements are available, but the logic and UI is in place for this to also handle multiple via points, short platforms, request stops, and also integrating these features into dividing services, too.
I've attached a couple of examples from the site.
Fictional West Coastway service with divide, request stops and short platforms:
View attachment 145574
CrossCountry to Brighton with multiple via points:
View attachment 145575
Virgin Pendolino to Edinburgh with short platform:
View attachment 145576
I am pretty sure that he has since changed it to used the Amey script on his site.
The script that Worldline (Atos) and KeTech use is:
"x is a request stop and passengers for this station should ask the conductor on the train to arrange for the train to stop to allow them to alight"
The script that Amey use is:
"customers may request to stop at x by contacting the conductor onboard the train"
This is the one that is currently on the Celia and Phil announcement generator and is the correct Amey script.