How to handle multi-language subs?

[hector]

Well there are already some discussions about this:
viewtopic.php?f=1&t=3094
viewtopic.php?f=1&t=14982
viewtopic.php?f=1&t=14203

But don't worry, this is not about modifying OS. Or perhaps it is but I'd like it to be more of a general discussion. How to handle these issues when you find them while you are subtitling some film.

Here is an example of an extreme case:
http://www.imdb.com/title/tt0120802
http://www.opensubtitles.org/en/search/ ... movie-5557

This is a multilingual film. As I said it is not very common but very often you can find some words in a foreign language (not the main language of the film). What to do?

Should I choose a single target language for the subs and translate everything to that language? I think this is how most people do it. What if I can't understand some languages in the film? And if I choose this approach, should I include the parts of the audio track in that language?

Some people perhaps would like to have the original script in every language. This is my case. Well, I know this is not very common. I'm aware I'm the 0.00004 % (let me round down). Then you have the problem that in OS every sub has only one language. It's OK because this is the general case and it's enough for almost all subs. And how does this interact with the possibly future "literal" flag? What does literal mean in a multi-language film?

I'm just looking for some answers, ideas, opinions... just what I think a "forum" should be: a place for discussion, not just a place to use as a problem-solver for one particular problem you have at one particular moment. And I'm not saying that it shouldn't be.

[YuriNorstein]

-deleted-

[hector]

Maybe my main concern is when I'm doing not a translation but a transcription, i.e. writing down the dialogue. One of my clearest examples is A los que aman (1998). It is a Spanish film. The main language is Spanish but has some dialogues in French. If I'm doing a transcription (the target language is Spanish) the first thing to decide is whether to write those parts in Spanish or French. The usual thing to do is to write them in Spanish. But in this case I would like to have both scripts in its original language because I can read (perhaps not hear) both. But as you say it would be a problem for OS because there is only one language for the subtitle. Then the obvious thing to do would be set it to the main language, Spanish in this case. But then you can't differentiate Spanish-Spanish from Spanish-French.

Perhaps the first thing would be to know if the subtitle is a translation or a transcription. Now you can't tell just from subtitle language information. You have to know the main language of the film. Another issue here is what about dubbing? Perhaps you have a dubbed version and you would like to find matching subs for that version. Then you see that most subs don't match exactly the audio. See viewtopic.php?f=1&t=15083&p=32084&hilit=literal#p32084

All of this could be summarised as "How do you use subtitles?" or "What do you want them for?" I usually like them to match the audio because I like to learn the language. I know this is not the usual case. Anyway it is obvious that multi-language subs are poorly supported.

[SmallBrother]

Maybe for this there should be a 'language' called "original" or "transcript".

After all, it doesnt matter what language it is, it matters that it's the same as the spoken language. Then it also doesn't matter if the spoken language is one, or a mix of two or seven languages. Obviously (I think) the Spanish parts should be in Spanish, the French parts in French, etc.

Like this it's also clear that it's not a subtitle- because most probably it will violate all possible technical criteria such as line lenght, durations, cps, etc. In this form it will be useless as subtitle, but valuable as source for making subtitles.

And this is what hector means/wants, right?

[hector]

Maybe for this there should be a 'language' called "original" or "transcript".

With a flag and everything? It could be a good approach.

But you are excluding the cases where the subtitle matches some dubbing. Take for example an English film with English audio. Here in Spain it is usual to have the Spanish dubbing (I don't know whether for good or bad). If I want Spanish subs for this film I could take two approaches: either take the English dialog and translate it or take the Spanish dub and make a transcription. The latter could not be labelled "original" but it could be labelled "transcript". Though not for the English (original) audio it would still be a transcription (of the Spanish audio). Perhaps there are not many people concerned about non-literal subtitles but I am. It is very uncomfortable for learning purposes to have the text not matching the audio.

So, I would see "original" and "transcript" as two different things. "Original" could be just another (pseudo) language and "transcript" would be a flag added to the subtitle record (it would still have its own language).

Yes, I know I said this was not about changing things but then what's the point of discussing things? :-)

Like this it's also clear that it's not a subtitle- because most probably it will violate all possible technical criteria

Why should a multi-language sub violate that? Yes, maximum CPS would probably be violated if you are writing every word literally, but not necessarily line length, durations, etc.

[hector]

About the "original" pseudo language I was thinking...

Yes, I know I'm too picky about these things. Usually people are not interested or pay attention to little details. But I like to have more information. In today's world information is power. Why can we see all those non-disclosure agreements and all that? Well, that's another story. I defend open standards and free information flow. And I like to have access to every detail.

If you use the "original" language you have to know the language(s) of the film. Where to look?

And I think it is funny how IMDB handles languages. You have the field "languages". Then for a particular film you can see in this field: "English, Spanish, German, Russian". Then you watch the film and that means English with 5 Spanish words, 10 Russian words and I'm still wondering about German. So, the point is that you don't know the quantities. If you see "Spanish, French" it could be 50-50 or 99.9- 0.1. You just don't know.

[YuriNorstein]

-deleted-

[SmallBrother]

Very nice.
But to me it looks like 'overlapping'.
And it could make players vomit (?)

Kinda off-topic: this could be also used to display the name of someone talking, in for example a documentary or so.
I am wondering if this is compatible enough...

[hector]

Okay, I meant both languages in the same file, not at the same time.

I agree with SmallBrother. This seems to me like an extension by some player. But I repeat: if someone knows where is the formal standard definition of an SRT, please, let me know.

I think this is like HTML some years ago. Everyone uses its own extensions. But they are not standardised. For HTML at least you have a more or less well defined standard by an organisation.

[YuriNorstein]

-deleted-

[hector]

Well I just tested it in the player I use and it doesn't work. And yes, it comes from SSA. But I think if you want to use it you better write an SSA instead of an SRT.

The fact that it works on some players doesn't mean it is standard and it can be used in everyday subtitling.

I still remember the HTML case. Currently it's been corrected but about 15 years ago an HTML page was displayed differently depending on the browser you used because every browser interpreted it in its own way. There must be a well defined standard.
Anything else should be taken as a private extension. Great for you if it works but not good for sharing or distribution.

But thanks anyway.

[YuriNorstein]

-deleted-

[hector]

MPlayer SVN-r31918

It doesn't work because SRT was never meant to write or understand those SSA commands. Some players implement it because they are too permissive. But you must be aware this is non-standard. Which is to say it is not guaranteed that it works everywhere.

[hector]

I work in IT and I know what I am talking about.

Imagine that you want to show in your subtitle the string "{\an item to be reviewed}'. I agree it is not a common string to show in a subtitle but nothing prevents you from doing it. So you write your SRT with this string to be showed and suddenly you find out that some software doesn't take it as a string (as you intended) but as a command. You are out of luck. You don't even know what this innocent string could do. It depends on how it's being interpreted. That's why standards are so important.

[YuriNorstein]

-deleted-