#pentagram@irc.freenode.net logs for 5 Dec 2015 (GMT)

Archive Today Yesterday Tomorrow
Pentagram homepage

[00:08:55] <-- nutron has left IRC (Ping timeout: 260 seconds)
[00:11:10] --> nutron has joined #pentagram
[02:49:07] --> DominusExult has joined #pentagram
[02:49:07] --- ChanServ gives channel operator status to DominusExult
[02:49:10] <-- Dominus has left IRC (Ping timeout: 246 seconds)
[02:49:27] --- DominusExult is now known as Dominus
[11:48:12] --> LaZar0 has joined #pentagram
[12:36:35] <-- LaZar0 has left IRC ()
[14:18:12] --> LaZar0 has joined #pentagram
[15:13:42] <wjp> Best advice I can give on LPC would be looking it up in a textbook or such
[15:14:09] <wjp> But it'll be ... intricate, probably
[15:46:14] <LaZar0> Hi
[15:46:23] <LaZar0> I've searched on the net about LPC
[15:46:58] <LaZar0> It is a simple algorithm, but I think that everyone can modify it for the needed compression
[15:47:54] <LaZar0> With U8, the problem I've found is that it has been (probably) modified only for this game, and if you don't know the intrinsics... it will be a bit difficult and time consuming find some positive result
[15:48:10] <LaZar0> Apart, how can I found the factors numbers?
[15:48:22] <LaZar0> It should follow some type of pattern...
[15:48:56] <LaZar0> but it must be exclusive for U8... and not having the source (nor the binary) tool that U8 people used to compress the original voices...
[15:49:06] <LaZar0> it will be very difficult to find something that works.
[15:50:11] <LaZar0> Even the factors created by sonarc.exe are different from u8 voice files
[15:50:35] <LaZar0> I've seen some old IRC log with Colourless that talked about it
[15:50:55] <LaZar0> and he do one tool to decompress with sonarcx the original U8 files
[15:51:09] <LaZar0> I've arrived to the same conclusion... they can be recognized but are not the same
[15:51:28] <LaZar0> So, the algorithm has been changed from the original Sonarc algorithm, definitely
[15:52:11] <LaZar0> Oh... and another counterpart is to know how is calculated the checksum to verify it is a U8 voice enconded...
[15:52:31] <LaZar0> But I think that that could be discovered with the own U8 voice files though...
[15:52:58] <Dominus> hmm, can't you work backward from the way we decode it?
[15:53:15] <LaZar0> I've tried to do that
[15:53:26] <LaZar0> but my first "stop" is calculating the factors numbers...
[15:53:34] <LaZar0> how are they calculated?
[15:53:45] <LaZar0> Inversing... I assume it would be:
[15:53:53] <LaZar0> - First do LPC, after that do EC compression
[15:54:01] <LaZar0> inverse the original dcmpsound, that is
[15:54:10] <LaZar0> But for LPC, I need to obtain the factor numbers...
[15:54:19] <Dominus> I may misunderstand because I'm really bad at coding, but if we have means to decode it, then we should be able to encode it, right?
[15:54:49] <LaZar0> Of course, you're totally right... but let's say this:
[15:55:10] <LaZar0> I use a table (123, 153, -13, 84) and a factorial algorithm to calculate something...
[15:55:17] <LaZar0> Can I inverse it given a result?
[15:55:32] <LaZar0> ... without knowing the code nor the table
[15:55:36] <LaZar0> ... of course
[15:55:44] <LaZar0> I think that this is what is happening with LPC...
[15:57:30] <LaZar0> Maybe I'm forgetting something or misunderstanding the dmpcsound algorithm...
[15:57:40] <LaZar0> I'm not so good in programming... ;)
[21:01:38] <wjp> It may require some university level knowledge of mathematics and signal processing
[21:03:23] <LaZar0> Yep, I thought that... I can program, but this level of mathematics surpases me...
[21:03:44] <LaZar0> I will leave the voices in english... the subtitles will be in Spanish but if Sound FX option is checked they are not visible
[21:04:01] <LaZar0> And there is no option to hear Voices with subtitles... it is a shame
[21:04:20] <LaZar0> Maybe hexediting u8.exe can be done this... but I don't want to break the original code
[21:04:29] <LaZar0> APART of this...
[21:04:42] <LaZar0> Doing my second betatesting with spanish dialogues I found:
[21:05:00] <LaZar0> -- After killing Khumash-Gor, when returning to speak to Zealans gods with obelisk...
[21:05:44] <LaZar0> If I get into the floor by the UPPER part of Apathas, very near of him... the dialogues are duplicated... do you know that?
[21:06:00] <LaZar0> If I get into the floor BELOW it, between Odion and him... all works well...
[21:06:14] <LaZar0> I'm asking if there is somehow to check where should be the bug...
[21:07:04] <wjp> Hm, that sounds familiar
[21:07:57] <wjp> Might be the first entry in docs/u8bugs.txt
[21:08:17] <LaZar0> I'll check it, I don't know there were anything like that...
[21:08:46] <wjp> If it's this one, and if you're recompiling usecode anyway, you can easily fix it
[21:09:35] <LaZar0> yeah, I now I've seen it!!
[21:09:46] <LaZar0> In fact, it will be the second bug I'm fixing
[21:10:12] <LaZar0> The other is the zero '0' shape that shows when Bane or Vardion are telling us the points in the pentacle
[21:15:01] <LaZar0> Have in mind that I'm not recompiling all the code... only decompile and compile those needed instructions (offsets, jumps, spawns...) that need some type of change...
[22:10:34] <LaZar0> Oh, and I also restored the text content in Resurrection Book of Teurgists. It shows the Intervention text of class 0x577, is not the corresponding for this book.
[22:12:06] <LaZar0> Class 0x0265: 1A64|0577:1D26
[22:12:07] <LaZar0> Class 0x0120: 16B2|0577:1D26