The Ur-Quan Masters Discussion Forum

Hello,

Does anybody think it would be worthwhile to sit down with the PC version and try to extract all the text?

Is it encrypted or compressed?

Has anybody tried this approach so far?

No, I'm not volunteering (yet), but it seems like from reading a few posts in this forum that many people would like the option of playing UQM with no speech and *all* of the original PC dialog.  I don't think it's worth playing through all of the PC version to get it though -- way too much work.

A second idea -- which is distinct from the first one -- it's talked about how there aren't existing recordings for some of the missing text in UQM.  What about just forcing the game to display text and have zero audio for these sections?  Wouldn't most people rather have this than miss the story altogether?

I looked in the forum a bit for this but couldn't find it -- I apologize if this is frequently brought up.

Cheers,
Chris

It is something that will need to be done.  Extracting the data from the package files is pretty much the only way to get this done; turning it into a format that can be understood by the communications code is a separate task.

Due to the nature of the way "constructed" lines work right now, if the PC version doesn't have "blanks" where the current script does, we'd have to wait until the string system is redone.

--Michael

This has apparently already been done (http://www.classicgaming.com/starcontrol/sc2/quotes/). The text in UQM has been adapted to fit the 3DO version (the 3DO version apparently didn't have any text), as it is harder to change the speech than the text.

Unfortunately, the 3DO version seems to have a few omissions that people have been complaining about. These missing lines of dialogue could easily be added to the text, as you suggest, but the sudden disappearance of spoken dialogue would probably result in even more complaints.

The best quick fix is to add the PC text as an alternative version of the dialogue. I believe this will be implemented once the new resource system is complete. In the long term, rerecording the speech may be an option.

Actually, I got all of the text files from here:

http://www.sa-matra.net/quotes/index.html

Tsing even mentions there that some of the Yehat text is missing.  Also pointed out to me is that some of the Melnorme text is missing, which I have confirmed.

Also on that page, he mentions Mudry, which was head of the Free Star Control Project.  Their source/programs for extracting this text might be found there?  Truthfully, I have no idea....

http://fscp.sourceforge.net/

http://sourceforge.net/projects/fscp

That website also mentions a new repository beside the sourceforge one...

http://infosrv.tech.klte.hu/svn/mudry/fscp

We've got the PC files. fOSSiL has extracted them with modified versions of Mudrony's scripts. Mudrony had the problem of not having access to the source, and he didn't know everything about the package format.
We have the 3DO source, from which I deduced the 3DO format, but the PC version was different in some places (eg. it used 16 bits values instead of 32 bits values), so fOSSiL still some more work to do. Also, he added code for extracting the lz compressed parts.
But now, we've got everything.

You might find this (http://www.customc.net/staff/sage/uqm_txt/) useful. These're the text files I extracted from the UQM packages.

meep-eep, can I get a copy of those files?  It bothers me that the ones on PNF are incomplete  :-/  If you are feeling generous, please e-mail to starcon@classicgaming.com

So, it sounds like meep-eep or fOSSiL has the relevant files, but the text still isn't fully extracted / decoded / whatever.

Correct?

I'm very glad to hear that somebody is working on this.  I want to play through the full game again (the last time was years ago), but I'd really rather wait until the text matches the PC version word for word -- just for old-times sake, you know?  I'm sure you know.

Anyway, great work guys, it seems I tell people about UQM on a  daily basis.

Cheers,
Chris

Chris: We were starting out with the PC texts and had them match the 3DO voice (the 3DO had no text of its own). When we've got the new resource system, you'll be able to choose. For now, we haven't bothered with the PC texts themselves.

Chad: I don't actually have fOSSiLs modified scripts nor the unpacked files now, but he's going to send me to them and I'll forward the files you want to you.

The PC comm texts are partially extracted. That is you can look at them (in hex viewer) but they are not in any kind of useable format yet. From what I noticed, the 3DO content (not just text) was somewhat mellowed down as compared to PC version -- must be the marketing and shooting for desired rating  ;) .

I, personally, cannot see why so many people want *the* PC script though, other than the very notable exceptions: Melnorme MetaChron thing and the fact that nobody in 3DO ever mentions Supox having the Ultron. The later I suspect was a bad edit of the speech track. If you listen to utwig033.ogg, you can hear the pause where they should tell you about Supox+Ultron.

And regarding the no-speech segments, that is exactly what UQM has right now in some places -- text only and no speech track.

... And Talana doesn't say where Syra is, and the Mycon don't say some important things...

How was it 'toned down'?

i vote for 2) no speech-segments with text not recorded as voice

I cannot remember all of them, of course, but for example, when Pkunk Fury regenerates energy Pkunk shout an insult. Well, PC -> 3DO, "nerd" and "nitwit" were removed and "dummy" added. The betatesters probably did not like being called "nerds"  ;)
There are more example like that.

That was the idea. But it's probably going to have to wait until we have the new resource system.

well, i mean ... really for the ultimate sc2 game you want to be able to take the best parts from PC version and the 3D0 version.  which for me would be pc text + plus pc "not toned-down" content, combined with 3D0 graphics and possibly videos.  quite frankly i can live without the voices.

of course, in a truly ultimate restoration the user could pick and choose between pc and 3D0 stuff, and then you wouldn't have any silly debates like, "it should be *this* way, not *that* way" ... you could just choose which you wanted, possibly even all done at runtime.

... just keep coding, just keep coding, that's what we do, we code, we code ...

cheers,
chris

Our vision looks like your "truly ultimate version."  Note the vast number of command line options that can be mapped to "=pc" or "=3do".

You guys really should make the PC dialogue the default in v0.4. Newbies who didn't play the original *really* need to hear the Melnorme metachron thing, and also the Mycon homeworld stuff, and a lot of people aren't going to bother looking at command line operatives. If you're worried about the inconsistency with speech, then you could turn the speech off by default (I personally would rather hear the music than the voices anyway).  

On the subject of the missing speech, a possibility occurs to me. How about taking apart existing speech files in the game and constructing the missing dialog out of them?

Great idea.  Let us know when you are finished with that.

I ain't a sound engineer. And the only speech text I got is the UQM ones which don't have what we want. I'm sure someone here can do a better job than me anyway.

I think what he really meant to say was, "If you just began to try you would see how utterly hopeless the endeavor is, because speech loses its fluid realistic quality when chopped up into little pieces."

In other words, we should be able to put together the missing Mycon dialogue quite well. :)

The rule of thumb back in my A/V days was "1 hour on the edit deck = 1 minute of finished footage"  From my meager experience splicing audio, the same rule basically applies, with the ratio of work to results growing exponentially depending on the complexity of the task.  Contructing decent-sounding new material out of old (finished) material is quite a complex task.

So, if we need to "reconstruct" 10 minutes of audio (a number I just made up), you are proposing a minimum of 10 hours' work.  On top of that, you have just stated that you are not willing/able to do the necessary work yourself.  As such, I put comments like yours in the same bin as "someone should really solve the whole Israel/Palestine conflict"

Yes, but someone should really solve the Israel/Palestine conflict! As I have no experience myself, I'll just head over and post that on their forums...

one problem with trying to piece together the audio is the echo/reverb on the mycon voice - it would be quite difficult to get that to sound right if we're chopping up words. Unfortunately, looking through the mycon subtitles, it doesn't appear that there's enough whole words to splice together for the missing dialogue. Without the echo, someone might be able to make a phenome or diphone db sufficient to (probably poorly) synthesize the missing lines, but with the echo's I'm sure it will sound horrible.

We could design a voice-mimicing neural net text reader, and train it on the current mycon speech and dialogue.  Then feed it the missing text, and voila!  Hey, maybe I _should_ solve the middle east's problems!  ::)

You know, the spooky thing is that I'm never sure when C20 is serious about these things. After all he IS some sort of Tech Wiz...

Yikes. Training it to be good enough would take an eternity.

I was thinking more along the lines of making a voice for festival using the festvox tools, but labeling would be a pain, we probably don't have enough recordings, and like I said there's all the effects.

We'd probably be better off just to record someone else saying the lines, and try to use the echo to hide the fact that it's a different voice.  Maybe add some noise and say there's some electrical storm or something interfering with the signal?

I wonder how much PR3's voice has changed since he recorded the Mycon lines. If it's close enough, we should have a realistic chance of fixing the Mycon voice with only a little effort (assuming we can convince him to record the missing lines).

As spurk and RibRdb pointed out, I was joking this time.  It's kind of like saying it'd be easier to get into space if we just built space elevators; the tech's all available, but the time and manpower involved would be _far_ more than just getting the original voice actors, or a good approximation.

If I were to get such a system perfected, I think the NSA, CIA, and numerous other countries' Intelligence agencies would want to have a chat with me (behind closed doors; a lot of them).  Such tech would have much abusive potential.

festival would be an option if the voice actors weren't around, but festival's voices aren't easy to get right (although you can fool them into saying things that sound the way you want by misspelling them: linux -> linucks, iPAQ->eye pack).  You could specifically tweak a voice to sound good for the two or three sentences that you need, and terrible on everything else.

Novus probably has the best option of them all: get PR3 to say the line about 20-30 times, then send the sound files to the Music remix team, who I assume have the sound editing tools that can make the voice a perfect fit with the originals.

Looking through the voice actors, it seems that many of the missing lines (Mycon and Dnyarri) are by PR3. The other most annoying omissions are in the Utwig (Greg Johnson) and Syreen (Madeleine Wild) lines. We may be able to track them down via PR3 or FF and record their missing lines. That would allow us to fix the most commonly criticised problems in UQM neatly, as long as we can persuade these people to help us out a bit and we manage to process the audio so that it sounds like the original recordings.