[PLEASE DELETE]

[Guest]

[PLEASE DELETE]

Edited March 11 by AlphaTwentyThree
[PLEASE DELETE]

[Puterboy1]

I think SAN also applies to Star Wars: Making Magic CD-ROM.

[Guest]

Huh, really nobody? Quite a surprise to me actually if I'm being honest, given that those games are 20+ years old already. You would think that the format is well-known by now. 🤔

[DKDave]

On 3/10/2024 at 8:38 AM, AlphaTwentyThree said:

Huh, really nobody? Quite a surprise to me actually if I'm being honest, given that those games are 20+ years old already. You would think that the format is well-known by now. 🤔

Yeah, the format is known.  It has some slight compression/encoding, but there doesn't seem to be any decoders I could find.  I've been experimenting with writing my own.  Still not complete yet, but here's a example of some decoded audio so far.  It sounds correct, but I need to do some more testing.

 

refintro.zip

[Guest]

Oh, didn't know it's actually its own format, I thought I've seen the .san extension in multiple games over the decades.

Thanks a lot for your work on this, pretty nifty! Dumb question: if you get it working safely, wouldn't it also be pretty easy to add to vgmstream? 🤔

Also, I'd be interested in hearing if my script correctly pulls out the audio since that might be used as a basis to play the movie's audio natively as well as for playing the non-demuxed videos (like with bik and thp).

[DKDave]

It's not really it's own audio format, just part of the SAN videos, so I'm not sure if they would add it to Vgmstream.  For IACT, it's just PCM16 with some slight encoding, so I've just saved them as .wav files.  I don't think there's a header you can put on the raw encoded data to make it play, not that I've found.

To pull out the initial audio, it combines all of the IACT blocks first - there are 0x1a bytes of header and then the audio block data in each IACT chunk.  It's odd because some of these audio blocks start in one IACT chunk and continue in the next.

I've attached my Python script for this one.  It works on all of the Droidworks SAN files.  There are other games with these SAN files that use different audio types, so I might add support for those as well.

 

 

san.zip

[Puterboy1]

Now for a Python script that can decode AFC files.

[Guest]

On 3/13/2024 at 8:41 PM, DKDave said:

It's not really it's own audio format, just part of the SAN videos, so I'm not sure if they would add it to Vgmstream.  For IACT, it's just PCM16 with some slight encoding, so I've just saved them as .wav files.  I don't think there's a header you can put on the raw encoded data to make it play, not that I've found.

To pull out the initial audio, it combines all of the IACT blocks first - there are 0x1a bytes of header and then the audio block data in each IACT chunk.  It's odd because some of these audio blocks start in one IACT chunk and continue in the next.

I've attached my Python script for this one.  It works on all of the Droidworks SAN files.  There are other games with these SAN files that use different audio types, so I might add support for those as well.

 

 

san.zip 1.44 kB · 2 downloads

This is pretty amazing I have to say, thanks a lot! If you take a look at what my script does, how would that need to change to properly "demux" san videos?

What I'm thinking about doing:

• get the demuxer to a state where it includes all the data from the SAN files needed to decode the audio, i.e. kind of the "pure" audio data
• check other SAN files and potentially expand the demuxer
• when enough variants are gathered, suggest support to be added for vgmstream with the nice information you already gathered

Looking at older formats like THP, those "demuxers" just remove the video portion from the file and leave the extension as is for a decoder (or player) to work with them. The issue is that I'm completely unfamiliar with Python (yet), so I'm pretty lost in any code. If you still have the motivation, I'd be happy to get a quick run-down of the format so I can adjust my demuxer accordingly. The goal would be a SAN file that only includes the audio data, so it makes sense to just keep the IACT headers intact I guess.

Happy to get a DM here or if you contact me on Discord (`alpha_twentythree`)

[Puterboy1]

Have you tried working this with vgmstream?