Teletext time travel 

7 January 2016

Transdiffusioner Jason Robertson has a complicated but fun project underway – recovering old teletext data from VHS cassettes.

Previously, it was possible – difficult but possible – to recover teletext from SVHS recordings, but they’re as rare as hen’s teeth as the format never really caught on. The data was captured by ordinary VHS but was never clear enough to get anything but a very few correct characters in amongst a massive amount of nonsense.

Technology is changing that. The continuing boom in processor power means it’s now possible to feed 15 minutes of smudged VHS teletext data into a computer and have it relentlessly compare the pages as they flick by at the top of the picture, choosing to hold characters that are the same on multiple viewing (as they’re likely to be right) and keep trying for clearer information for characters that frequently change (as they’re likely to be wrong).

All of this takes phenomenal amounts of processing power and therefore time. The longer you run the data comparison, the better the results – but we’re still talking days rather than hours in many cases. Moore’s law should see this fall in future, but until then Jason is ploughing on – and making the results available to us all online.

The pages are a snapshot of life in the 1980s – British Rail train times, Mrs Thatcher’s opinions, new pound coins and Gus Honeybun – and therefore fascinating for historians of modern life as well as the huge numbers of us who love – and miss – having teletext at our fingertips.

Unique Code & Data [Jason Robertson]

You Say

24 responses to this article

Flanagan 8 January 2016 at 1:56 am

Very interesting. Could it be applied to Philips N1500/1700 recordings from the mid-to-late 70s? They’d arguably be even more fascinating.

Russ J Graham 8 January 2016 at 1:38 pm

Replying to a similar question on Twitter, Jason said “Probably! If you can source a Hauppauge WinTV PCI card, the software works with that.” and linked to this article on his blog:

Martin Jarvis 9 January 2016 at 12:33 am

I’ve still got my Grundig VHS recorder with its ‘plug-in’ teletext adapter which allowed me record ‘live’ teletext.

Jason Robertson 9 January 2016 at 5:25 pm

@Flanagan I have an N1700 and some tapes too, but it needs a service. Does anyone know anyone with suitable skills?

I have some pages from 1976 but they’re very noisy as they went from N1700 -> VHS. But they are indeed fascinating! Pre-1976 would be interesting too as the specifications were different, and different between Ceefax and ORACLE.

Dave H 17 January 2016 at 11:15 am

How about some crowd decoding effort on this?
You, Jason, define a preferred medium or format for people to record telext-off-a-VHS recording into using software we could access easily (VLC?).
– then perhaps upload them to a few One-drive accounts using some naming convention.
– Let people use your decoder to have a go at decoding them. (maybe try to not have too many people per file at a time in somehow – using open source booking out project software ?).
– then re-upload best results to a site for you to add to a server?

Just a thought..

Alistair Buxton 18 January 2016 at 4:11 am

Hi, I’m the original developer of the software Jason is using to do this.

There’s a few problems that make it quite tricky to spread out the work:

First you need special hardware to capture the raw signal, as normal teletext decoders cannot handle the degraded signal and will just ignore it. You also need special drivers for the special hardware. In practice this means you have to use Linux to do the capture part, although the decoder itself should run fine anywhere.

Second the raw data is huge. 2048 bytes per line, up to 32 lines per frame, 25 frames per second. A 3 hour VHS produces about 5 gigabytes of raw data. At least on my connection this would take almost as long to upload as it would to process it locally.

Third the raw signal will vary depending on the VCRs used to record and play it and also on the tape itself, and the software has to be tuned to account for this in order to get the best results.

If these issues can be overcome then distributed decoding is definitely something I would like to implement. In the long run I think there’s enough old VHS tapes out there that we could build an archive of almost every teletext page ever broadcast in the UK, given enough processing power.

Another thing I would like to do is use GPU computing to accelerate it, but I haven’t yet found the time to learn how to do that.

If anyone is interested in how the software works there’s a quite detailed explanation on the github page.

Stephen Donaghy 18 January 2016 at 10:53 am

Alistair I am oddly fascinated this. I presume that

Is the github repo you’re speaking of?

I am no good with python but I’m curious if I can do something to improve the process. I might look into it in my spare time.

Sadly I don’t have ANY VHS’s lying around so I’ll use the example data you’ve uploaded for now.

Kit Davies 18 January 2016 at 12:08 pm

20ish years ago I had a small decoder PCI card that accepted UHF input, filtered out the Teletext pages and pulled them in as data for saving & analysis. Don’t know if such cards still exist, but playing the tapes through one should be possible IWHT.

John 18 January 2016 at 3:10 pm

This is a very similar idea to the Hinman Collator, which was used to determine similarities between early printed texts (such as Shakespeare quartos).

richard broadhurst 18 January 2016 at 5:35 pm

Would it make sense to share (static) pages and confidence of each character so that someone else with the same page could add to existing results?

Michael Biel 18 January 2016 at 7:54 pm

I am in the United States but I have about 20 VHS tapes I recorded in London in the summer of 1983. I was never able to find any player, TV, or adapter for the Teletext and have always wanted to be able to read them. I have three PAL VHS machines here, and the machines and tapes are all in pristine condition because they have had little use. I also have many tapes I recorded in Austria of Austrian, German, and Swiss TV in 1989 and 1998. I suppose those tapes would need a different decoder.

Alistair Buxton 19 January 2016 at 8:02 pm

@Stephen that’s the one, yes. Don’t take it as a good example of how to write Python code.

@Kit all that is required is a capture card that can oversample the VBI lines in the video signal at a relatively high rate and then pass out the raw samples without attempting to decode them. Old Hauppauge WinTV PCI cards are capable of this. Other may be too. People have sent me data captured with other devices and it mostly works.

@John interesting. This is also known as Hamming distance in CS. That’s only the second step though. Before we can even try to compare pages the signal has to be deconvolved, and that’s by far the slowest part.

@Richard if two people had separate VHS recordings of the same pages then that would probably help. If they were both working from the same same tape but with different VHS machines then it might help a little bit. If they were both working from the same digital recording they would get identical results.

@Michael all the European (PAL) Teletext systems are compatible. The decoder software would also in theory work with NTSC or SECAM captures by changing only two constants, but I’ve never seen any such recordings as they are very rare.

Francis Kim 24 January 2016 at 10:49 am

Boy, this brings back memories!

John Veness 27 January 2016 at 9:27 am

@Alistair Buxton: “In the long run I think there’s enough old VHS tapes out there that we could build an archive of almost every teletext page ever broadcast in the UK, given enough processing power.”

This is a great idea – I’d love to see a publically accessible archive of teletext. I worry, though, about how many VHS tapes there really are “out there”. I expect only a small percentage of people still have any.

If Internet bandwidth is a barrier to sending 5GB files around, maybe snail mail of posting VHS tapes to one or more people who have the means to decode would be a good idea. It would be great if someone could start this as a proper project.

Eelco 23 February 2016 at 6:54 pm

Hi Alistair,
Can you tell me exactly what type of Hauppage WinTV card you use? Many thanks in advance.

Dan Farrimond 1 March 2016 at 4:25 pm

I am more than happy to help the recovery effort! Would like to see a VHS donation drive and some funding/PR backing from a public organisation.

Anyone willing to help me make it happen? :-)

John Veness 19 March 2016 at 4:29 pm

Anyone up for starting a website/facebook page for this? (Sorry, not me).

I have a few VHS tapes I could donate. Actually, some of them may be SVHS, as I remember my VHS recorder could play back teletext subtitles and teletext pages later (turn off and on-able, so not “burnt” into the image). I had loads more tapes but got rid of them years ago, which is what worries me about there not being that many left.

I see people getting rid of pre-recorded tapes on Freecycle/Freegle sometimes, or wanted adverts could be posted to those. I’d be happy to collect some of these tapes and post them off to someone to capture/decode, but I couldn’t store them myself as my wife is trying to declutter.

John Veness 19 March 2016 at 4:39 pm

Another thought: Are there any other projects out there where people collect old recorded-off-the-telly video tapes? I know there are people that like finding old adverts, or trailers, or idents, or continuity stuff, etc. Maybe there’s scope for combining efforts with these people?

John Veness 21 March 2016 at 2:49 pm

Me again. I’ve just picked up six tapes off Freecycle. To where should I post them? Is it just Jason who is doing this?

Graham 10 September 2016 at 10:34 am

I used the BBC computer and the BBC teletext adaptor..probably a PCI card to decode store and then print out the results for students at college. Mainly news and current affairs items. We used it in the Languages dept and was mainly channels like RAI UNO TVE etc which had if i am correct adopted our system by then. All delivered by satellite.

Graham 10 September 2016 at 10:39 am

I did decode teletext pages off video tapes but i found it hard to correctly decode a whole page without errors as the tape systems transport mechanisms seem to be slightly unstable for perfect speed playback. The best results were with Phillips VC 1500 series machines and Panasonic SVHS recorders. Hope that helps!

Michael 23 July 2017 at 8:18 am

I have an S-VHS tape of ORACLE’s last night and Teletext’s first. Please let me know if you want it to extract.

Graham Pearson 29 August 2018 at 3:11 am

I remember watching Ceefax In Vision sequences during the early 1980s on BBC1 and BBC2. If generated from the continuity booth there would be title cards and if generated from Hitachi television there would be page x00.

Phil Hale 29 August 2018 at 10:50 am

Followed the link here from the project. I too used to record TV on SVHS tape for the superior picture quality and discovered by accident back then that teletext data was pretty much preserved fine compared to VHS recordings – it quite a novelty to be able to relive the news of the day when watching the recordings. I still have these tapes (covering most of the nineties) waiting for the day to be captured into a PC – mainly for the off air TV gems that have never made it to DVD or have subsequently been edited for DVD release. I look at doing this exercise every 5 years or so figuring that the process would get easier as technology improves and through the collected wisdom of others who are also doing the same. Of course I’m also mindful that the source tapes are not going to last forever so I know I should get on with it! To be able to extract the Teletext data too is a real bonus and I’d be happy to add to the pages already captured. Do get in touch if I can help.

Your comment

Enter it below

A member of the Transdiffusion Broadcasting System
Liverpool, Saturday 28 January 2023