Recovering Teletext data from VHS recordings

[Read the post]


That hundreds, even thousands of frames of each teletext page are required to OCR each one is also a powerful tribute to just how astoundingly awful VHS is.

VHS was never intended as an archival format, which some people are now more acutely aware of, since this is now its sole purpose.

In the late 80s, I was trying to buy a reasonably-priced 3/4" VCR, such as a U-Matic for a few years, but failed. There simply wasn’t thousands of dollars more of tech inside to justify the expense.

Being in the US, I always geeked out on Ceefax. I would have entirely used it here.


I mean this in the most sincere way

That hundreds, even thousands of frames of each teletext page are required to OCR each one is also a powerful tribute to just how astoundingly awful VHS is.

To be fair, this is for recovering data transmitted in the overscan margins of the TV signal - outside of the TV picture - not the TV picture itself.

Use the entire TV signal for data and it’s a different story. Back in the 1980s Alpha Microsystems computers used VHS tapes for software distribution and backups.


I don’t think that’s what’s going on here. Isn’t this transmitted as digital pulses, like closed captioning?

1 Like

Teletext ist still up and running in germany…


twelve million users a day…


not exactly exactly - the figure is rounded : P

1 Like

And audio cassette tapes are terrible for storing computer programs. And it turns out that a pizza makes a lousy bicycle, and beer is a lousy engine lubricant.

VHS was perfectly fine for what it was, if you were using it for what it was intended.


VHS was manageably lousy for what it was.

Thought. The signal could be digitized using common RTL-SDR rig, feeding the baseband video through a video modulator and into the USB dongle. Then it is the matter of standard signal pipeline.

Working around the modulation and feeding the ADC with raw baseband may work better, though. Those 8-bit 1-channel 24 Msps USB oscilloscopes based on the FX2 chip could also be a good candidate for the analog front-end.

1 Like

I think maybe stinkinbadgers’ point was that it’s not surprising that such heroic measures have to used to recover teletext from VHS, because to look at the VHS standard (as opposed to SVHS) you’d be surprised to see it had captured even a tiny crappy bit of the space where teletext was stored.

Sort of like zooming in on someone’s eyeball in a photograph and maybe getting a glimpse of what the photographer looked like. Theoretically maybe possible, especially if you had a bunch of frames in quick succession, but probably requiring enormous amounts of processor-intensive manipulation to even come close.

Before spending the summer of 1983 busing tables to save up for a floppy drive, I used audio cassettes to store games on my Apple II+ clone.

Loading a program took many attempts, adjusting the volume each time, even after I started monitoring audio levels it with oscilloscope I got at a garage sale.

I even used an 8-track recorder - those briefly existed - because it stored a better quality signal than my cheap cassette recorder. I had Sublogic Flight Simulator - which later became Microsoft Flight Simulator - on 8-track.


band name!


Yes. It’s data sent during “unused” parts of the video scan. It’s collected up, and put into memory, where it’s displayed in the same way that a video card works in a computer. It couldn’t happen until digital electronics made it viable in a small package.

It sounds like they re having problems sync’ing to the digital signal. But I can use the closed caption on old videotapes (well I have one where it’s delayed, and incomplete), so I’m not sure this would be a consistent problem. But maybe it’s because these were home recordings, so something is marginal.

It also sounds like they re using a sledgehammer for this, digitizing it all, then decoding in the computer. It would take a bit of wiring, making a sync decoder to get the actual portion of the scan with the data might be easiest, then they’d just have to decode the data in that portion of the scan.

1 Like

and behold… it’s on the internet (at least for the 1st public program):

And VHS wasn’t intended to record the whole of the the PAL/NTSC signal either, just needed to be good enough to view the pictures. The subtitles/closed caption data could be included on several frames a second, meanwhile the several hundred or thousand other pages may only occur once every while (main and section indexes get repeated more often).

It wasn’t uncommon to see corrupt frames from broadcast tv, and I suspect FastText would take several tries to cache a page, if it didn’t see it right first time. So no surprise at all that ancient VHS tape might take a few tries to get an accurate page, and if you’ve got half an hour of footage, then why not take a look at all the versions of a page in that time to figure out a byte-perfect page. Sounds like maybe the archivists have devised some super-clever way to avoid any manual intervention, but I doubt it’s the cusp of anything.

1 Like

So how long was the 8track tape? If your program didn’t fit in one track, things would stall as the head moved to the next track.

8track recorders never seemed to sell well. I can’t imagine they were cheaper than a reasonable cassette deck. I used a surplus cassette transport, in effect a tape deck without cover. It cost about forty dollars. I added a power supply, volume controls (for both input and output) and level meters. I never had a problem.

I never had any purpose-made recordable 8-track tapes. Instead I was recording over someone else’s country/western music tapes. For the good of humanity. Each program took only 14 minutes (typically, to store all of memory from 0800-BFFF, the end of RAM on the Apple II series) of space. I could put one program on each track.

Sadly, my country/western eradication program did not extend to cassette tapes. I was stuck in a Volvo station wagon for two days, driving from Toronto to Winnipeg in the middle of winter, with three smokers and their country-western Christmas carol collection. Some times I still wake up screaming.


That’s a funny coincidence. Sometimes when I want to listen to some ambient music, I slap on a cassette of 8-bit software!