Speech Audio Format Has been Decoded

[DMJC]

Hi, I worked with Mario HCL Brito and we got the Strike Commander Audio format figured out. It's the same audio format as Wing Commander Privateer used. This is Mario's notes:

What we have here starts with a typical WC1/2 era header, which points to a bunch of entries: first begins at offset 0x1c, second at 0x602a, third at 0xdbd4, etc
Each of these entries is LZW compressed. I happen to have a Matlab script that can decompress these babies from my FMV cut-scene research a few years back ;) Really textbook LZW tho: start reading 9 bits at a time, expect symbol 257 as a stop symbol (although only some games honour that), and dictionary entries start on symbol 258. As dictionary expands towards entry 512, read 10 bytes at a time, and so on. You read at most 12 bytes at time, the dictionary doesn't grow beyond that (4096 entries... with the first 256 entries being the ASCII table itself). Oh and symbol 256 resets the dictionary.
Unpack these and you get a Creative VOC file, which you'll be able to play in VLC, or a similar player. The format is PCM, which means it's not compressed by any predictive audio compression algorithm. It's just plain 8-bit (unsigned), 11025Hz, mono audio. Good old 90s tech.

[remileonard]

Hi,

Thank you very much for theses informations. I am currently working on loading the data from the image of the GOG version (to have all the voices). So far i've mannaged to read the raw cd image and access all the files.

But the PAKs files in the cd version are differents from the floppy disks. The files are LZW compressed. I have started to implement the decompression using your comment, but i have a bug :) so here is a question, when we get the symbol 256, we need to reset the dictionnary, does that mean that we have to restart with 9 bits reads or should we keep the current bit width ?

Rémi.

[DMJC]

Have you been on the wcnews.com discord? There's a guy on there. Alltinker he's made an application called Originator which decodes these formats. He can help you with all the format decoders.

…

On Mon, 7 Apr 2025, 8:05 am remileonard, ***@***.***> wrote: Hi, Thank you very much for theses informations. I am currently working on loading the data from the image of the GOG version (to have all the voices). So far i've mannaged to read the raw cd image and access all the files. But the PAKs files in the cd version are differents from the floppy disks. The files are LZW compressed. I have started to implement the decompression using your comment, but i have a bug :) so here is a question, when we get the symbol 256, we need to reset the dictionnary, does that mean that we have to restart with 9 bits reads or should we keep the current bit width ? Rémi. — Reply to this email directly, view it on GitHub <#4 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAFS2Y56YUSHRI3YYT4MTX32YGQIZAVCNFSM6AAAAABZO6CWNGVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTENZUGM4DKOA> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[remileonard]

Well happy news, i've managed to fix my bug, the story, i wasn't reading enough bit to decode the symbol. The bug is fixed, next speaking dialogues :)