Railway automated announcement simulator

This is a little tool designed to simulate automated announcements as formerly found at many railway stations across the South of England, and currently found at a few in the North and many in Scotland.

Please note that this project is not affiliated with Realtime Trains, KeTech, or any other company in the rail industry.

It uses the Realtime Trains API to fetch realtime data. Note that this API is not perfectly suited for this purpose, due mostly to limitations in the source data, and so some functionality that you might expect such as train formations, catering information, and a few others things is missing. This is not something I can do anything about unless either the RTT API starts providing this information (which it mostly can't for technical reasons), or I switch to a different API.

It is designed for use with sound files available from elsewhere, which include Phil Sayer, Celia Drummond, and Alison McKay.

I have based announcement scripts mostly from memory with a bit of creative licence here and there, with the odd use of real recordings online to get details and timing right in a few places. However, I expect there to be errors, and I absolutely welcome pull requests if you're able, or issues raised if you're not. I really want to get this as accurate as possible in the long term! While the files are from KeTech I most fondly remember the Amey systems, so accuracy towards those is the primary target, but I am very happy to include options to simulate other systems I might not have remembered or considered, as long as the sound clips are available.

This script is still quite new and I expect a few bugs to still be present. If you experience a crash or an announcement that sounds wrong, feel free to file an issue (please include the output of the script if possible). However please do check the configuration hints down below.

I'm afraid I've not made getting it set up particularly easy or friendly, so I would rate this as only suitable for those comfortable with using a command-line. If anyone wishes to package this up into a more user-friendly format they are welcome to do so provided they comply with the terms of the (very permissive) MIT licence, and I would kindly request they let me know they are doing so.

Setup instructions

• Sign up for the Realtime Trains Pull API at https://api.rtt.io/ and obtain an API key and password. Note you should not need to request the detailed API for this, as I don't believe I use any fields from it, but reports from those who have tested it are welcome.
• Ensure that you have Python (3.11 or newer) and git installed.
• Clone both this repository and the repository containing the sound files into a convenient location:

git clone https://github.com/Muzer/rtt-announce
git clone https://github.com/Rail-Announcements/ketech-llpa-announcements

• Change into the directory of the repo containing the sound files.
• Modify the rename_snippets.py config section in your favourite text editor — change the output_dir (for instance, to renamed_wavs/) and change filetype to wav. Be sure to save it.
• Now, run rename_snippets.py, eg:

python rename_snippets.py

• ...this will create a directory containing wav files renamed appropriately for my announcement script.
• Now, edit the rtt-announce.toml file you cloned from this repository. Fill in the API key and password, and fill in the full path to your renamed_wavs directory you created above. Tweak any settings you might want to change.
• Install python dependencies (sorry, I have no idea how to package Python stuff nowadays):

pip install pyaudio requests soundfile

• Finally, you should be able to run the script:

python rtt-announce.py

• You can pass a different station as a command-line argument, but any other settings changes will need to be made in the config file.

Note that when the script starts running, there will often be a large "backlog" of delay, cancellation, "now standing", and safety announcements to get through. However at small to medium stations this will calm down fairly quickly. At large stations it might take longer to "catch up" to the present time; you may wish to consider disabling some announcements or parts of announcements, or making delay announcements less frequent — see below.

Configuration hints

• You can choose from voices "Male1" (Phil Sayer), "Female1" (Celia Drummond), and "Female2" (Alison McKay). Note that the latter has a lot of missing voice clips so features will be limited, and it will not work well at all at stations outside of Scotland.
• Almost all announcement types can be turned on or off individually. Don't like so many announcements about arrivals? Switch them off! Hate the safety announcements? They're gone too! Check out the announcements_enabled section for this.
• Not hearing any announcements, or only hearing announcements for certain trains? Check on the Realtime Trains website and see if these trains show as "Approaching", "Arriving", and "At Platform" at appropriate times. If not, it is likely that this area has been resignalled relatively recently and Realtime Trains does not yet have the new detailed timing information to produce this data. If this is the case, I suggest you try enabling departures_trust_triggered and arrivals_trust_triggered announcement types. This is a degraded mode of operation so the full functionality will not be available, but this will allow you to hear a single announcement which will be triggered whenever a train's arrival time or platform is confirmed.
• Announcements too long but don't want to turn any off entirely? Most announcements can be customised to reduce the length or frequency, and many iterations were once found across the rail network so a lot of variants will remain authentic. As an example, you can disable the calling pattern from any announcement, you can disable repetition of the introduction on "next train" announcements, you can reduce the frequency of delay announcements or disable new announcements whenever a train's delay changes, and many more.
• For demonstration purposes I have set up the default config file to have some features from both Southern and South Western stations. For instance from Southern land I have enabled "X, this is X" announcements for trains from elsewhere now standing at the platform, and I have enabled "train now standing" wording — but all of this can be customised and hopefully you should be able to easily craft announcements authentic to any region based solely on editing the config file. If there is an option you think is missing which would help you get closer to the authentic announcement style you remember, I will be happy to add it!
• By default announcements will not play if the train's destination is unavailable; however they will play if only a calling point is unavailable. This affects trains to places such as Tweedbank and Headbolt Lane. By default as well, if a sound clip (usually a calling point) is not found, it will repeat the previous sound clip — this is an authentic feature of the Amey system formerly used across the South.

Roadmap

I am now reasonably happy with the way announcements sound and the general script; however as mentioned above improvements and suggestions for improvements or new features are always absolutely welcome!

While I think I've got the timing of announcements fairly good, there are still some areas that don't quite sound right to me, and so I might in future do some tweaking to adjust the length of the small pauses between snippets.

In future I would like to make a truly authentic system with configurable channels of different voices for different platforms. If and when I perform this revamp I also plan for it to include authentic features like unimportant (eg safety) announcements being interrupted by train announcements, and then restarting once the important announcement is done.