VHS-Decode Project Could Help Archival Efforts

Archiving data from old storage media can be a highly complex process. It can be as simple as putting a disk in an old drive and reading out the contents. These days, though, the state of the art is more complex, with advanced techniques helping to recover the most data possible. The VHS-Decode project is an effort to improve the archiving of old analog video tapes.

The project is a fork of the LaserDisc-focused ld-decode, started by [Chad Page] back in 2013, which readers may recall was used for the Domesday Duplicator — a device aimed to recover data from the BBC’s ancient Domesday LaserDiscs. VHS-Decode is designed to capture the raw RF signals straight out of a tape head, which are the most direct representation of the signals on the physical media. From there, these signals can be processed in various ways to best recover the original audio and video tracks. It’s much the same technique as is used by floppy disk recovery tools like the FluxEngine.

Despite the VHS name, the code currently works with several tape formats. VHS, S-VHS and U-Matic are supported in PAL and NTSC formats, while Betamax, Video8 and High8 tape capture remains a work in progress. Using the code requires a video tape player with test points or traces that make signals from the head accessible. Capturing those signals is achieved via a Domesday Duplicator hardware device, or alternatively a Conexant CX2388x analog-to-digital converter, often found in many old PCI TV tuner cards. Various techniques can then be used to turn the captured signals into watchable video files.

We love a good archival project, and VHS-Decode is clearly a useful tool when it comes to salvaging old video tapes.

[Thanks to JohnU for the tip!]

47 thoughts on “VHS-Decode Project Could Help Archival Efforts

    1. It is named after a laser disk community action project done by the BBC to commemorate the anniversary of the original British doomsday book. This project was called the new doomsday book

      The records were pressed onto laser discs, but sadly those disks are failing due to faulty pressing equipment, so the doomsday duplicator was built to archive them. This was then used to develop the ld decode software.

      1. A version of ld-decode existed first, then it was pretty much rewritten to support the Duplicator and do PAL Laserdiscs better. Then more people came in and brought a CD/digital audio decoder (same thing on LD), a port of the PAL Transform Decoder, and most recently AC3 audio support.

    2. Unfortunately some of the information in this article is quite out of date. Those additional formats mentioned have great support now, and the video comparison was posted in June of 2021. Below is a more recent (but not current) to date comparison with much better results.


  1. Wow, I’m really impressed by the uplift in quality in that sample video. Feels like what a good S-VHS recorder could do on a new tape back in the day.

    My level of tolerance for glitchy video has never been high, and that goes from completely unwatchable to “huh, not bad!”

      1. Honestly we wished this outdated video wasn’t demonsteated here. There are new comparisons out here. Fortunately the original signal capture of this tape is still available in our archives so we can try decoding it again using new version of VHS-Decode…

        1. Why not? You are still on the same place where you was 5 years ago, quality is same bad quality as before 5 years and you, the group of deep amateurs cant do it better. So 1,2 or 5 years before, the quality and reliability is the same – BAD. Instead, on the professional forum on November was demonstrated fully working system with better RF capture quality and professional decoding written by signal processing programmers.

  2. This video is 1 year outdated, decode has evolved a LOT more since this was made.

    RF Capture and preservation is the initial process that’s 1:1 preservation, its hardware agnostic and it covers all formats, this writer did not read the wiki preserving the media has been done for years.

    Betamax, Video8 & High8 have decoding support just not as well refined as VHS & SVHS that’s the reality of today.

  3. I worked on a project in the 1990s that used a high quality VHS tape and a specialized tape mechanism as a device to store streaming high speed data. At the time we couldn’t get fast enough sustained write speeds using more traditional methods like hard disks or tape drives. I wonder if this project can read VHS data tapes as well (not that I have any, just idle curiosity)?

    1. Read off the magnetic data for later analysis? Absolutely!

      As for how to get any understandable data out it in the nexr step, not out of the box.

      Depending on the used data format, that could range from “not too bad” all the way to “incredibly hard”.

      But at least the flux data could be saved for future analysis that way. :-)

  4. If the project limited to hijacking a ‘stock’ high quality VHS machine to capture the head output, or could you excise the head and transport from a cheaper machine and modify it to ‘under-run’ (possibly using a replacement motor if needed for speed stability/control) to read out the tape at a slower speed and higher quality?

      1. VHS requires a coupled tape feed rate and head scan rate, but the requirement for a specific scanout speed is only enforced by the need to directly display the output on a TV. If you do not have that requirement, then you can decrease tape speed and scanout speed as long as the two remain in sync to ensure correct tracking.

        The question is not if this is possible (it is), but whether it would be worthwhile doing (i.e. whether head quality is a significant enough factor that reducing data rate would increase data quality).

    1. Yes, you could, but there is no need to. The signal to noise ratio is fine anyways.

      But since vhsdecode taps directly from the head, you could absolutely do it. However, parsing the analog data would be much more complicated, since you would get overlap. I don’t think it’s worth the trouble.

      The reason for using SVHS decks is that they are usually more stable in their speed. (Mechanically.) But there are VHS decks which are just as fine. Most of the circuitry on the player is bypassed and ignored, so a fancy player with great onboard processing is not inherently better for this.

      Another thing which is interesting about vhs-decode is that once you have the saved “rip” with the analog raw data, you can use improved future versions of vhs-decode to get even better decoding as it improves.

      In contrast, a great transfer with traditional Time Base Corrector and “3D filter” can be really good, but the result is “baked in”. You can’t go back and improve it further.

      For archival purposes or for cramming the absolute best out of a VHS tape, vhs-decode is excellent in my book.

      Also it brings great rips “to the masses”, before you had to have really unusual, expensive studio level equipment to do a great job.

    2. That may be theoretically possible but the head drum needs to move at the appropriate speed to properly track the tape. And the software parameters like sample rate would need to be adjusted.

  5. I’m wondering if one could use an SDR like the HackRF to capture the signals from the heads. I think it would work with the video heads but not 100% sure about the audio since I haven’t messed with VHS/NTSC in decades now.

    1. Iys NOT possible by many reasons. There are no capture drivers for this SDRs, they are 8 bit and do not have enough bandwidth for this special purpose. As well, they decode to IQ which is not the case with VCR decoding.

      1. Yes, HackTV and FL2k based software can transmit both baseband and RF modulated video and audio currently using certain HackRF and certain USB to VGA/DVI dongles. Beware Sulio Pulev, he’s an obvious troll.

    2. Yes not only is that possible but it has also been done. 9954tony and I think a couple of others have tried with a high bandwidth SDR (you only need 8MHz of bandwidth for VHS). But the signal to noise ratio is really no better than a Domesday Duplicator capture device or Conexant PCI(-E) card.

    3. What they’re using *is* like one of those… except substantially higher quality than the HackRF, to get a good result, and direct-sampling because there’s no use (and plenty of downside) to having a 0-6GHz tuner in front of the ADC for this application.

  6. For the past week I’ve been mentally working the idea of storing data on VHS. Yes I know it’s been done many times before, but I’m thinking more of novel ways to achieve it rather than what’s been done already. My current idea is to convert files into a series of QR codes and save each one as a frame (or two) on a VHS video stream. I have a barcode scanner that can read level 24 qr codes at about 15 per second off of a BW CRT.

    However, after reading this article, I’m now contemplating writing data directly to the tape by connecting directly to the heads. I’m thinking a QAM64 signal piped out of my sound card at 192khz directly into the read write heads. I have some software that can achieve 400kbs with the signal described above and the ability to decode the same signal later. Have to think about the logistics of pulling it off but that might be my next project.

    1. It’s been bugging me since yesterday — I can’t remember the name of that VHS data recording system. I do remember it had a memorable name, ironically. I’m pretty sure I’ll recognize it when I see it. So far all my search results are not fruitful. I’m gonna bug some of the old engineers on that project and see if they remember.

      I do remember that eventually we were able to get SCSI RAID arrays that would handle our data. And the old VHS “disk drive” were retired.

      1. There was also a consumer system called something like “video backup” which used (maybe) a parallel dongle to connect to a composite video input on a VCR.

        I remember seeing ads for it in the Amiga magazines in the 1990s, but I never saw one in person.

    2. What you’re “planning” is a very low-res version of what Sony did in the early 80s. Betamax/VHS decks were paired with PCM adapters which converted analog audio to digital, and the digital signal converted to NTSC or PAL video, as a pattern of black and white blocks (very small blocks! Way smaller than a QR code!) On playback, the reverse process converted the sequence of patterns back into high quality audio. The first CDs were mastered on such systems, based around U-matic video recorders.

    3. Jack, PLEASE tell me that sometime in the last 5 months, you’ve abandoned that idea of storing data on video tape! Video tape of ANY type is a fragile medium subject to damage and destruction. Why use tape when reliable, high-capacity and CHEAP hard drives are available, as well as USB docking bays to allow the big drives (I use 3TB at the present) to be stored offline? I realize this forum is intended for the reporting of creative ideas. But please channel your creativity into something safe and useful. Storing digital data that is important enough to be kept safely should not use tape for storage, unless no other medium existed.

        1. Creating a mechanism that has high risk of losing important data needlessly isn’t play. Just don’t let yourself or anyone else who has important data USE your new toy. If the tape is damaged (which you likely wouldn’t find out til NEEDING the data years later), then the guilt for losing the data falls fully on you for inventing the dangerous tool. But if you like wasting your time on inventing a harmful tool, have at it. But most of us have lots of non-harmful endeavors that really need solving. THOSE items are the “best play”.

          1. There are millions of master recordings of commercial music releases from the late 80s to about the 2000s stored on S-VHS.

            ADAT tapes are S-VHS tapes with PCM encoded digital audio stored on them. I say until the 2000s because that is when most digital audio recorders became hard disk based.

            Hopefully most of them have been copied digitally to other media, lossesly, so that the data in the copies is bit identical to the data on the original tapes.

  7. You know, I always wondered why when archiving old footage from professional videotape formats, we record from the composite video out from the VTR, and not the tape head directly, which is outputting the FM signal.

    I mean, composite out is demodulated from this fm signal, usually using older analog components to do the demod when you could be directly recording this high frequency FM signal to digital, storing it (RAM, HDD SSD) and digitally demodulating the compsoite signal from the FM, using software or hardware.

    I mean an FM signal is easy to convert to digital, because it is simply the length of a waveform which represents the video data. When you put modulated FM carriers through a limiter they also become frequency modulated square waves which are either on or off for certain lengths of time, so it’s binary which can easily be input into a computer. The computer just needs a timing reference to compare how fast the ons and offs are happening from the FM signal.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.