SS2 subtitles

[6648934cbe01aZylonBane]

You sometimes forgot to remove the stutter, punctuation is wrong in some cases and so on.

Or I intentionally left the stutter in and the punctuation is correct.

I've already added the reaver and SHODAN subtitles and will be posting them later. Rumbler and assassin barks have no spoken words so it seems like a waste of time, but I'll probably do them eventually.

[6648934cbe33eMarvin]

That would be inconsistent and when I'm talking about wrong punctuation, I mean missing punctuation marks. Yes, even you are able to make mistakes.

Ok, then I don't have to do it myself. BTW, would you be interested in a script which automatically creates (non-timed) subtitle files from vlogs and emails or will you really not do them, ever?

[6648934cbe444ZylonBane]

If someone provided pre-generated sub files for the logs and emails, I can't think of any reason why I wouldn't use them.

[6648934cbe530Marvin]

Ok, will send you a PM when ready (next weekend probably).

[6648934cbe627ZylonBane]

Here's the current alpha snapshot, with SHODAN and psi reaver voices added, and various other tweaks.

EDIT: Removed attachment. Current version in first post.

[6648934cbe733Marvin]

Thanks. I don't get it: The German version provides extra files for the verbal psi reaver sounds but all of them are the same as the english ones. Oh well.

Edit: Which program do you use for synching text and speech?

[6648934cbe808ZylonBane]

SoundForge.

[6648934cbe8eaMarvin]

Ok, that's a tad beyond my budget.

[6648934cbeb3cMooztik]

If someone provided pre-generated sub files for the logs and emails, I can't think of any reason why I wouldn't use them.

i ve made them for french translation
in sub and timed srt files

[6648934cbec8bMooztik]

i can extract -> convert english text from .str files to .srt (with generic timing)  and post them here

and convert .srt files into one .sub file

//update
ZylonBane : i formated the sub files like yours

[6648934cbedcfMarvin]

That would be a big help, don't have to do it myself then.

I'd be grateful if you could do that for the German sounds and texts. You can get the *.str files here (the SCP version, it has the most up-to-date level0x.str files). I'm assuming, of course, you're doing this with an automated script.

[6648934cbeec5Mooztik]

ok, fixing errors in my parser and i send theses files...

[6648934cbeff4Mooztik]

done !

if you find any errors in the text structure, let me kown it. everything is made automaticly. i will fix it.

i've added tabulations to the sub file. then, by using column mode in ultra edit we can copy /paste text or timing to another translated SUB file with same structure.

if you have timed srt files for other subtitles file (other than emails/logs), i can generate sub files.

[6648934cbf21bMarvin]

Cool. So how are you automatically determining these "generic" timings? String splitting at punctuation marks and setting a default length per character?

One probem is that the German files are encoded in OEM850 as that is the codepage Irrational used for special characters such as umlauts or sharp s (ä,ö,ü,ß,Ä,Ö,Ü). Those are missing in the German files, but that's easy to fix.

[6648934cbf4c7ZylonBane]

Ok, that's a tad beyond my budget.

All I use it for is viewing start points and duration. Any free audio editing software would do.

[6648934cbf8c7Mooztik]

Cool. So how are you automatically determining these "generic" timings? String splitting at punctuation marks and setting a default length per character?

One probem is that the German files are encoded in OEM850 as that is the codepage Irrational used for special characters such as umlauts or sharp s (ä,ö,ü,ß,Ä,Ö,Ü). Those are missing in the German files, but that's easy to fix.

yes generic timing is based on poinctuation (. ? ) and lenght from number of characters * 0.53 (medium time to read a character)
but, there is no regularity in theses audiofiles. absolutly horrible if used ingame as is

i can change the format .i used it for french files. what format do you need ?

i'm using  'visual subsync' . it's a wysiwig video subtitle creator software. but you can use it with audio files (some tweaks to do).

[6648934cbfc1bMarvin]

All I use it for is viewing start points and duration. Any free audio editing software would do.

I'm using Audacity which is a little bit finnicky. Works, but could be better.

@Mooztik: Your text parser would need to translate the file to something more manageable (UTF-8, ISO 8859) first and then start reading. When I'm editing it, I can reconvert to OEM850 and everything's okay, but if you read the file with, say, UTF-8 and save it in that codepage, all special characters are garbled up forever and I would have to correct everything.
It would be best if you could fix that somehow and don't split the strings. Just put everything in one multisub, I have to go through all files manually anyway and then decide how many parts are necessary.

Visual subsync, I'll remember that, thanks.

[6648934cbfd4fMooztik]

php act so strangly with caracter encoding... for real it sucks.
i don't have Locale for dutch, then php do nothing with theses strings...
this is why strings keep there format to oem. i can't fix it from inside the parser. need to use Ultraedit.

when you say, 'don't split the strings'  whould you want a  'sub(text)' or a 'multisub{}' with only one line ?

[6648934cbfe77Marvin]

A multisub with a timing of 00 to some arbitrary large number, 100000 or something like that.

[6648934cbff70Mooztik]

like that ?

[6648934cc00b4Marvin]

Exactly like that, thanks, you're a champ. 

[6648934cc01f0Marvin]

Is there any way to switch to a different description within the same multisub block such as when Polito/Shodan is interrupted by Xerxes or the Many?

On a different note: How the fuck are you supposed to subtitle the Many sequences with the interlocking sentences?

[6648934cc03e2Mooztik]

Is there any way to switch to a different description within the same multisub block such as when Polito/Shodan is interrupted by Xerxes or the Many?

I think, no.
actually i removed all the description lines. most of the time we know who is speaking, and when there's more than one caracter, i write the other one uppercases. for exemple xerxes  '<<XERXES IS SPEAKING >>' . purely for aesthetical display.
but, without description, it's possible to simply put the caracter's name before the texte on a new line.

i will surely make two versions for the french subs. one corresponding to the main release by ZylonBane, and another one without any description and some more text cut timings for better french sentences timing.

[6648934cc07d1Marvin]

You can disable the descriptions in cam.cfg with a simple switch so there's no need to delete them from the files, more is always better in this case.


Concerning the emails in general: I've taken a lot more liberty with them than with all the other barks or conversations. In all cases except The Many's one, the emails are almost 100% infodumps. I think that the player will care more about having the information displayed in the most concise and practical manner possible, so I've purposely stuck some sentences together which are separated by a pause if they provide correlated information and I've separated lines which come immediately after each other when they have different topics.
This is easy in Polito's case but Shodan makes so many long pauses, mostly due to stuttering and echo, that her subs are almost exclusively one line per sentence. Then come the Many: They mostly taunt (or horrify) the player, so their subs are more cutscene-like, separated whenever a pause happens, even mid-sentence. Example:

Code: [Select]

multisub EM0503 {
{ time 2300 length 4400 text "Misstrauen ist die Tyrannei des Individuums." }
{ time 7600 length 4150 text "Deine eigene Gattung betrachtet dich als eine Bedrohung." }
{ time 12350 length 3050 text "Warum ermordest du unsere Einheit?" }
{ time 15900 length 3700 text "Egal. Es ist nun alles klar." }
{ time 19850 length 2600 text "Du wirst nicht mehr weiterexistieren." }
{ time 22550 length 4700 text "Es ist nur die Frage, wer dich eliminieren wird:" }
{ time 28600 length 800 text "Wir." }
{ time 29580 length 500 text "Oder." }
{ time 30400 length 1500 text "Du?" }
}
The last three words could easily belong in one sub, but you'd completely miss the dramatic effect. So, to summarize: Polito has longer subs due to infodumps, Shodan has mostly one sub per sentence and the Many has (have?) the shortest subs. This is inconsistent by design but I think fitting the respective speech pattern of the character is more important than consistency in this case. But please engage in a discussion if you feel different.

Another little problem is the interlocking speech of The Many and Shodan. Right now, the subs start when the first word of a sentence is spoken and stops at the last word, no matter if the word comes from an echo and is repeated with larger volume later. Also, especially with The Many, this leads to subs still going when the next is already starting (echo of former sentence is still going on, pre-echo of latter sentence is already beginning). However, the subtitle engine is configured per default to let subs stay longer than defined anyway (subtitles_extra_time <time>, with time set to 3000), so you would only notice if you changed that variable to something very small.

Another thing where I'm stumped is capitalization. Example:

Code: [Select]

{ time 14800 length 1700 text "Verschwinde hier... " }
{ time 17000 length 5500 text "oder wir werden dich so verwunden, wie du uns verwundet hast." }
OR:

Code: [Select]

{ time 14800 length 1700 text "Verschwinde hier... " }
{ time 17000 length 5500 text "Oder wir werden dich so verwunden, wie du uns verwundet hast." }
(same with commata, but with commata it's clearer).


Last but not least: I don't know if I'll ever come around to subbing the logs, but here's a cleaned up version of Mooztik's file without timing. Works, but ugly.