HOWTO: Read Dolby SR-D audio from sound negatives

[Tiago João Silva]

Hello.
As promised, I'm sharing how to read the 5.1 audio from SR-D sound negatives.
First, some words of warning:

• Sound Negatives shouldn't be played in an intermittent transport (like a film projector). A continuous transport (like a studio dubber) that can also play both base-in-front or emulsion-in-front is mandatory.
• Reading the SR-D 5.1 sound, be it from sound negative, married print, sound print, or from the MODisks themselves, even if someone ever finds out how to digitally decode the SR-D files in the MODisk to PCM (WAV), should be considered a last resort; since AFAIK nobody knows how to convert digitally (the code is locked inside the Zoran DSPs in the Dolby cards), this process involves a conversion from SR-D to analog and another analog to digital conversion.
• SR-D 5.1 sound was lossy compressed with a relative of the AC3 codec, and not only has codec artefacts but it will suffer even more with further AAC conversions down a broadcast or delivery chain.
• If there is multichannel analog or PCM sound of the final mix (be they AKAI DD8, ADAT, DTRS, Sony DASH, Mitsubishi ProDigi, WAV/AIFF/SD2 bounces), those are always a better source than the SR-D sound. Only use the SR-D sound if nothing better is available.

We are using:

• Sondor OMA-E film transport synched to video
• Kinoton Basement Reader (with flywheel) jury-rigged to the OMA-E's film path (on the right-hand side, after the film starts descending to the destination reel - bolt it to the side of the cabinet frame, about level with the headblock)
• SR-D_LoopFilmPath.jpg
• Dolby DA20 (CP500 would work as well, but with the DA20 we don't have to worry about defeating the B-chain settings). CP-650 wasn't tested.
• DOS-based PC with specific Dolby software (DRAS10.EXE, QC403.EXE) and serial connection to the DA20. We haven't tested if updated DOS versions, or the Windows equivalents, could work as well.
• The computer can be anything from a 486DX to hardware equivalents of the latest rack computers supplied by Dolby with CA10 sound camera adapters (Pentium III, 440BX chipset). We're using an HP Vectra VE i8, Pentium III/633
• Analog-to-Digital conversion, inputs in -10dBV=0VU sensitivity, trimmed with Cat.No.69T to -20dBFS

I'll attach the information supplied by Mr. Thomas G. Kodros (engineer at Dolby) to Cineric. It's pretty clear and I can't explain it better. Basically, install the supplied software, and use the command-line switches as indicated. In the end, it was pretty easy - after we had the information, that is. That's why I'm sharing.
Most SR-D positives are missing frames in the beginning and in the end, so the sound negative is a far better option.

Good luck and let's keep film sound alive.

[David Ferguson]

Thanks for the tips to switch the DA-20/CP500 to "reading negatives" mode - I had no idea they had the ability to do that. And also for the QC software, I didn't have a copy of that so I will add that to my archive.

It is also interesting to see the name "Mr. Thomas G. Kodros" - I had seen the acronym "TGK" in a few Dolby things before, but didn't know who it referred to.

this process involves a conversion from SR-D to analog and another analog to digital conversion [...]Dolby DA20 (CP500 would work as well, but with the DA20 we don't have to worry about defeating the B-chain settings). CP-650 wasn't tested.

Ideally you want a CP650 with the digital output card, because then you are not doing a digital -> analogue -> digital step. Or you can do what I have done, and modified by DA-20 to output digital audio, skipping the DAC. I currently have it outputting ADAT (throwback! Who remembers ADAT?) because that's the easiest for me to capture multichannel digital audio, but making it output AES3 pairs would be easy too, and would align with most current studio and cinema equipment. In fact, using a modified DA-20 is probably the best method, because then you're absolutely sure there is no additional processing going on in the CP500 or CP650 that you don't want.

or from the MODisks themselves, even if someone ever finds out how to digitally decode the SR-D files in the MODisk to PCM (WAV), should be considered a last resort; since AFAIK nobody knows how to convert digitally

Yeah the printmaster on the MO disc is also currently unknown, and it's not currently clear to me what, if any additional processing is done by the CA-10 when it reads the MO disc and records the SR-D negative. It's not nothing, because the two formats don't seem to resemble each other, but it's probably not too much either (because why would you not have that additional step done pre-MO disc?).

SR-D 5.1 sound was lossy compressed with a relative of the AC3 codec, and not only has codec artefacts [...] Only use the SR-D sound if nothing better is available.

Indeed. I love your machine and it sounds like it works well, but it also does concern me that it's necessary already. Have original audio masters already gone missing? It's only been just over 30 years!

supplied by Dolby with CA10 sound camera adapters

Slightly off topic, but do you have/have access to a CA10? That's one part of the Dolby SR-D process that I'm still trying to find info on - I have the CA10 manual, but haven't seen one in person (or actually, seen any pics of one) yet. I'd also love to know what software runs on the PC connected to the CA10 to control it, and read from the MO disc.

[Tiago João Silva]

Hi, David! Nice work with your SR-D wiki, I had browsed it myself already when I was looking for some way of reading SR-D negatives!

The original audio masters have a tendency to go missing when a country's film "industry" can't even be classified as a "manufacture". Portuguese films have always been living hand-to-mouth, and sometimes even eating their seed corn.
For some films, the sound negatives were lost
For many films, the magnetic final mix was lost, and only the optical SNeg remains - producers didn't care about the magnetic, the sound negative was the final product. Who cared if the magnetic remained in the lab that had struck the sound negative as collateral for unpaid bills? Safety copies? That's expensive... Or the producer folded, their warehouse was emptied and the magnetics were sent to the skip...
Even when printmasters started being DATs, the issue of being left forgotten on the labs and no safety copies remained.
So no wonder that very few producers ever considered spending the money on a backup of the 5.1 mix on a much more expensive format like ADAT or DTRS. The final product is the MODisk, they paid a pretty penny for the Dolby license already. And a clone of the MODisk? Why-ever for? The lab only needs one! And that's why for many films, the SR-D sound negative is our most original material for the 5.1 mix...

We have a CA10 (which we're also using to take the 5.1 + Dolby Stereo mixes off MODisks when nothing better is available) and I hope we can help you. Below the Nuoptix AOS on the attached picture, that's a CA10 "crate" without the front panel attached.
CA10+AOS.jpg

That one came to us without the original computer, which meant we had to get a computer, the Dolby ISA-bus interface card, and an Adaptec SCSI card to get it to work. Thankfully the MODisk drive was external and wasn't lost with the computer, one less thing to scour eBay for.


I thought the CA10 DOS software was in the Warehouse, but I'm not finding it, so Dolby must've sent it to me. It's too big to attach to this post.

Now, if anyone, ANYONE, has the MODisk that came with the CA10, that also didn't come with our CA10. On one side it had a backup copy of the CA10 software, and on the other side it was formatted as Dolby MODisk that could be loaded in the CA10 software and had a series of SR-D files that are necessary to set up and align the LED matrix that printed the SR-D blocks to the sound negative. We have blank MODisks to dump the raw files into, if someone has them, but Dolby UK was unable to find those files for us.
I actually can't attach most things Dolby sent me, they're too big to attach. Even just the PDF of the CA10 Installation Manual would give you a very good perspective of what makes the CA10 tick (e.g, a list and description of all the cards in the crate)

Two photos of the CA10 interface
CA10 playback.jpg
CA10 file browser.jpg

[Tiago João Silva]

David, I've sent you a package with technical information about the CA10 (installation/calibration/commissioning) and the Dolby software.
How hard would it be to change our DA20 to digital output (ADAT would be fine) and maybe even our CA10 too? (although there AES would be better)

[David Ferguson]

Thanks so much for all the info you sent, it has lots of useful information that I wasn't aware of already. For example, I had no idea that on the Dolby Printmaster MO, the analogue audio (the .SVA files) are encoded as Dolby AC2. I'll have to see if that helps me read anything useful from them.

It is also great to see some pictures of a CA10 - I've mentioned this to you, but to put it here as well, it is interesting to note that the CA10 and DA10 share quite a few cards.

For converting the DA20 to digital output, it is not a tricky technical thing to do, but you'd want a proper solution, rather than what I have which is some wires soldered onto my DA20 that then go to a prototype board with the ADAT circuitry (I'm sure I had a picture, but I can't find it right now). The best solution would be a drop-in replacement card for the 686 DAC to make it output AES3 (or some other digital output) - and a few people have now told me they'd be interested in such a thing, so maybe I should look into it. It shouldn't be hard, just need to find a suitable AES3 transmitter chip (I had a quick look and couldn't find a 6 channel one, but I can probably get away with 3 stereo ones). If I ever do that, I'll let you know.

[Min young Park]

Hi Tiago, great to connect online! I’m also working with CA10 and Sondor Resonance.

I was curious about scanning Dolby SRD sound negatives on the Sondor, I scanned only print film for Dolby SRD, but it seems like you've already mastered it—amazing work!
Could I ask you about Dolby SR? How are you scanning Dolby SR in Portugal with the Sondor?

[Tiago João Silva]

Obligatory: do you have the Dolby CA10 Alignment and Test Disk, by any chance? Do you have any way of copying the files inside?

As for reading SR-D sound negatives, don't thank me, thank the Dolby engineers that had the forethought of putting the code in to read sound negatives, which I was pretty sure they had done; why would they force themselves during the development of the system to take a print out of every test sound negative just for reading, when it was little more effort than flipping the bit values in the code? The extra code must have paid for itself in the first week! And then they shipped that debug code in production, and didn't forget about it later.

The Kinoton sound head would be able to read the SVA track as well, but not from negative. We read SVA tracks (positive and negative) just like you must be doing it too, with a Resonances (on a Sondor Versa) - but it needs a separate pass after the Dolby Digital track and sync both by the two-pop (greatly helped by Dolby mandating the pop at ACTUALLY two seconds from FFOA, because otherwise I just KNOW everybody would have continued to put the pop wherever).

The Czech Film Archive is getting a Scanity and it seems like they will be reading the optical sound from the full-frame scans - how exactly, I don't know, their visit here in Lisbon was a whirlwind and I couldn't get many details. I could have misunderstood and it could be just a regular reader or even Resonances too, since Scanity scans film in real-time.
Sondor/Université La Rochelle never finished that part of the Resonances software (scanning to images, and then converting from images to sound). By having an intermediate step of image files, you could even do manual repair/dustbust of the sound track image before converting to sound; Resonances remained a "stream" processor - it converts a stream of image sensor values to a stream of digital samples.

[Elia Orselli]

The Czech Film Archive is getting a Scanity and it seems like they will be reading the optical sound from the full-frame scans - how exactly, I don't know

maybe they are using AEO light or something derived from that tool? https://usc-imi.github.io/aeo-light/#hero

[Sean McKinnon]

would'nt a sound negative still read with an optical reader? Maybe a little noisy but I would think it would track? You may have to flip the phase or invert the track in your DAW.

[Scott Norwood]

With regard to reading SRD negatives: wouldn't it be easier to flip the polarity on the camera side? I don't know enough about CCD cameras to know how to do this, but it should not be hard to switch it from producing a negative image to a positive one.

[Tiago João Silva]

would'nt a sound negative still read with an optical reader? Maybe a little noisy but I would think it would track? You may have to flip the phase or invert the track in your DAW.

An analog sound negative can be read with a regular reader, with a lot more noise, because most of the noise comes from the transparent ("white") part of the track and even stuff like "noise reduction" bias on the optical track is trying to minimize the white area of the track on the positive when there's silence, which means maximum white area on the negative when there's silence; but most importantly, it will have more image-spread distortion.
This distortion is a by-product of the photographic process, but thankfully, it goes in the opposite direction from negatives to positives. This means that to get zero distortion in a positive, you have to introduce on purpose protographic distortion on the negative, in a specific amount for each lab/stock/printing light combination. Badly-compensated image-spread distortion can be heard as "spitting" on 'ss' and 'sch' sounds.
A regular reader can't compensate for that effect.
Many special readers have promised to compensate for this effect and avoid IS distortion; my only experiences have been with LaserInterface and Resonances, and I could never get good results from LaserInterface.
With Resonances, we can change the exposure and simulate changes in printing density and the sensitometric curve when reading negatives (up to a point) so we can get closer to minimal IS. Since we no longer have the cross-modulation tests for most of the sound negatives, we can go a lot faster by ear with Resonances than by printing a series of exposures to actual positive.
The other thing with Resonances is that it's mostly immune to film grain and less-than-perfect opacity of the black sections, so the background noise is a lot lower even when reading positives.
More information about Image Spread and Cross Modulation can be found in the Kodak H-44 document I'm attaching, "Cross Modulation Testing for the Motion Picture Laboratory"

[Tiago João Silva]

With regard to reading SRD negatives: wouldn't it be easier to flip the polarity on the camera side? I don't know enough about CCD cameras to know how to do this, but it should not be hard to switch it from producing a negative image to a positive one.

It's a matter of trying. Although I think that would also flip the perforation signal which I fear is being used for sync/clocking by the system.