Decoding AC3 tracks from film ?

[Mediahead]

Is it at all possible ? Has anybody looked into it? Google is not helpful for this

[stwd4nder2]

There was a thread on the OT forums where some people were trying to reverse engineer it. I can't find it now, and I don't think they made much progress on it.

[borisanddoris]

In theory, one may be able to accomplish it digitally since the track is actual AC3 at around 320kbps once it gets into the Dolby processor. The only output mechanism from that is analog out for connection to a cinema processor but if one was crafty enough to access that data at the right point, I think it might be able to be done, digitally. The DA10 used in house chips for decoding I believe, while the DA20 used Zoran chips. The key would probably be capturing the data signal before then. Not being a hardware engineer, I don't know the ins and outs of such things but again, in theory, it might be possible.

Shoot, even SDDS is some flavor of ATRAC, which was used in MiniDisc. Get the right brains on this, and anything is possible.

[dwalkerdon]

There is a Film Scanner that can decode Dolby AC3 from 35mm called Sondor Resonances. here is an article about it: https://dft-film.com/dft-installs-first-...ilmmuseum/

[davidferguson]

The older topic about this can be found at https://fanrestore.com/thread-2633.html, where I posted some of the technical info about AC3 and how it's done on film. However.... since posting that, I've received some new information which (for now at least) basically means it's impossible infeasible to rip the raw Dolby Digital data off the film. Below is a summary.

- Dolby Digital on film isn't actually AC3. It's a pre-cursor of what then became AC3, and as such is not compatible. I don't know what changed, and the people I know at Dolby wouldn't tell me. 

- Straight after the 76x76 data blocks are read from the film, they are XORed against a constant in order to break up large black or white areas which would be hard to detect. Because the same value is used during encoding as decoding, the original data is recovered. I don't know what this constant value is.

- After being XORed, the data is run through a concatenated code, which is where two reed-solomon decoding instances are applied one after each other. Also some of the bits are only protected by one of the reed-solomon layers, but I don't have any more information about which layer or which bits.

---

While it's therefore infeasible to rip the Dolby Digital off the film directly, what it is possible to do is use a Dolby Digital decoder to decode it to audio, and then record that. Ideally you want to record it when it's still digital, rather than having the decoder to a D->A and then immediately do a A->D again. This isn't terribly tricky if you know a little electronics and have a Dolby Digital film decoder you can play with.

This is also what the Sondor resonances is doing, it just has a standard Dolby Digital soundhead on the scanner, which is fed into a standard Dolby Digital film decoder. Nothing special!

I was actually doing this last week, recording the PCM digital straight from the Dolby Digital decoder (DA20), and got some great results with it. You can hear a couple of clips if you want, of the Dolby Aurora logo, and the THX Broadway logo, showing the comparison between the audio on YouTube, and the audio straight from the film. Listen to it on a surround sound setup if you can - the difference is amazing!

---

I've done a reasonable amount of research into this, so any questions, I can try and help with.

[BusterD]

A guy who normally reverse engineers retro video game stuff was working on Dolby Digital a while back, but I'm not sure if his work will be of any use since it's all over my head: http://furrtek.free.fr/?a=dolbyd

[captainsolo]

I have always assumed if you could capture the digital signal while the print is running it would then be like trying to decode DTS undecoded digital signals.

[davidferguson]

A guy who normally reverse engineers retro video game stuff was working on Dolby Digital a while back, but I'm not sure if his work will be of any use since it's all over my head: http://furrtek.free.fr/?a=dolbyd

Thanks for that link. I'd seen Furrkek mention this on twitter, but I hadn't seen the webpage until now. I've sent them some more information that may be useful about this.

I have always assumed if you could capture the digital signal while the print is running it would then be like trying to decode DTS undecoded digital signals.

You have a choice of where to capture it, either before the decoding, or after the decoding. Capturing it before decoding is (currently) useless as no software decoder currently exists for the variant of Dolby Digital on film. Capturing after decoding, when it's PCM audio, is very easy, and indeed is what I've done in the past to great success. See the links in my previous post for examples of that. However, both of these require having a full film setup, which isn't very common.

[dvdmike]

So it would be one PCM stream per channel?

[davidferguson]

So it would be one PCM stream per channel?

That's right. It's actually three stereo I2S channels, to make up the six channels total.