ANN: SubRip 1.50 b4

[ai4spam]

@SvenBent: Working on fixing it, but my motherboard died :(. Also, the correction is applied AFTER the punctuation correction, so feel free to use the dictionary to work wonders ;).
@iElectric: That's usually a bad/non-standard .sub file. Try a different source. Other than that, I don't know, Zuggy handles the .vob/.sub parsing stuff.

[SvenBent]

another "bug" (actualyl a limitation i think)

Error
Not enough memory to add this character to the Matrix!
Size of the Matrix in Bytes: 6326094

would you consider increaseing te maximum size for the matrix file ?

[ai4spam]

It's not memory that's the problem, it's the fragmentation. Now SubRip allocates each character on demand, when it should in fact allocate chunks of 10 or even 100, and fill them afterwards. Too many allocation requests eventually lead to the failures that you get.
So... as soon as my new motherboard gets in, I'll work on this as well.

[colonel32]

Hi, I'd like to report a bug. If you view/edit a character matrix which contains unicode characters (Korean), and try to "Delete This Character (Del)", SubRip 150b3 crashes with an "Access Violation at address 004C7AF8 in module 'SubRip.exe'. Read of address 00000000."

It seems to be reproduceable after a restart and a reboot. You can see yourself by editing the attached character matrix and trying to delete index 00094.

Hope this helps!

This is an excellent program, thanks so much :)

[ai4spam]

I'm working on a new version, which should take care of all these issues. I'll try your example when I'm done. Thanks for reporting.

PS: I'll repeat an earlier request: anyone wants to volunteer to design a new face for the SubRip page, once we move to SourceForge?

[iElectric]

I would gladly help, but only thing i am good at is Flash.

[ai4spam]

Well, a nice Flash intro wouldn't hurt either. No need for fancy menus or anything, just an intro would be cool.

[masken]

just like to say thanks for your continued efforts with this excellent and essential application ai4spam. Keep up the good work :)

[iElectric]

ai4spam, you have msn or something? Tell me more details about intro, i would like to help at making new home for subrip

[ai4spam]

Thanks for the offer. I have gMail (use the same id as here @ gmail). I don't log in often, but you can leave me messages that way. Just add my id to your contacts.

[SvenBent]

i have another bug.

Big I and litle i are detected as THE SAME even though they are Graphicaly different

I Reduced the matrixc size so that it did ask for the big I and tgen i entered some random character and the same character was used for the litte i

I can send you the subs ?

[LRN]

Maybe colors isn't set right?

[Teebeeke]

I was wondering if it could be possible to add a list of all characters to the dialog (as an option) , atleast a-z / A-Z / 0-9, to see which characters have allready been added. I always wonder how many i need to add before i have had all of the a-z/A-Z/0-9 characters.

[LRN]

It is useul only if you're using new matrix each time you OCRing.

[iElectric]

And option to auto-load the same matrix each startup would be perfect!

[ai4spam]

@Teebeeke: I'm also working on an improved version of the "fill matrix" algorithm, the one in 1.50 is working, barely.
@iElectric: I'll think about it.

[SvenBent]

I found more errors with the "format whole words only)

Before
41
00:07:26,360 --> 00:07:27,554
That<i>'</i>s all right.

42
00:07:35,160 --> 00:07:40,951
My name<i>'</i>s Gaius.
lt<i>'</i>s nice to meet you.

43
00:08:47,880 --> 00:08:49,552
lt<i>'</i>s Valhiss!


after:
1
00:07:26,360 --> 00:07:27,554
<i>That's</i> all right.

42
00:07:35,160 --> 00:07:40,951
My <i>name's</i> Gaius.
<i>It's</i> nice to meet you.

43
00:08:47,880 --> 00:08:49,552
<i>It's</i> Valhiss!

clearly something is wrong.
mabye you should make somw week letters that never change the whole word like ' and /

--- edit ---
Never write when you are tired :-)

[SvenBent]

i really would like to suggest a debug mode where the text is filled with the matrix index number instead of the characters.

That way you could easy track down a misunderstod matrix entry

--- edit ---
I really shouldn't write in forums when i tired

[ai4spam]

Clearly something is wrong with your keyboard :p.

Thanks for the feedback, I'll look into it some more. I've made some changes, but I'm still far from ready to release.

[Sinistral]

Hey ai4spam,

First off, thank you for the continueing work on SubRip. It's an amazing program.

Unforunately, I have two problems.

1: When some subs are ripped in SubRip 1.50b3, SubRip warns me that the timing is negative and that it might need a delay. However, most of the times all the timings in the sub turn negative and the sub is unusable (unless you know a way to positivate them).

I remembered seeing something about negative timings in one of your changelogs and I saw that SubRip 1.40b2 was the first to support negative timestamps. So I downloaded SubRip 1.40b1 and ripped the subs with negative timestamps, and most of them came out okay. So I thought you should delete the support for negative timestamps.

However some subs with negative timestamps, come out mostly okay, when ripped with 1.50b3, but when ripped with 1.40b1, they are f*cked (the timings are all wrong). So I thought it might be nice to have an option in which you can choose wether to ignore the negative timestamps, or to use them.

2: I've ripped a lot of subs with your program but I never came accros one I couldn't rip, until now. SubRip 1.50b3 only detects the top part of characters of this subtitle, but they're to small parts to make out what character it is. When I use the sub with a movie however, the subtitle is displayed fine. If you want the subtitle, you can download it here: http://rapidshare.de/files/32553152/Movie.rar

These are not very important things, so if you don't have the time to fix it, it doesn't matter too much. It has to stay fun for you too ;)

Anyway, thanks in advance for reading!

[masken]

Can someone please setup a working mirror? http://zuggy.wz.cz/ is just not downloadable and the mirror specified there has expired...

[DonGato]

What you can't get?
It's hosted at Sourceforge and you have plenty of mirrors.

[ai4spam]

@sinistral: You're probably talking about different kinds of negative timings. I don't really understand how some of them could have come out ok with 1.40b1. Since 1.40b2, the default delay that is added is so that the first sub starts at 00:00:00. That takes care of the negative timestamps on some DVDs that "count backwards". When you load subs with negative timestamps in vatious applications, you get various results: some ignore the sign, some wrap aroung (-1=255, etc.). The method I chose works out pretty well. Can you send me an example where it doesn't?
The subs you put online are deffinitely weird. The Chinese ones indeed only show the top part. I have no idea why. The English ones are completely blank (I gave up after a few blank images). Which ones do you mean?

@masken&DonGato: Sorry to hear about your download troubles. AfterDawn has a good mirror. Although SubRip has a project space reserved on SourceForge, we haven't begun using it yet.

I've been too busy lately for any progress (getting married). LRN has worked some on ttxt support.

[steelman]

@ai4spam: wz.cz is the worst webhosting in CZ. Its very slow server therefore people can't download from there or load the site.
btw congrats to getting married :D

[Sinistral]

Thanks for the reply. I got a subtitle here which turns out with negative timings if ripped by SubRip 1.50b3 and positive timings if ripped by 1.40b2: http://rapidshare.de/files/32833451/Negative_Timestamps.rar.html (I also included the character matrix for this subtitle).

And about the weird subs; my SubRip also detects the top of the Chinese characters and also the top of the English ones, though when SubRip detects them, they look like lines. You should change your Min. Char. Height to 1, maybe then you'll see them. If you have no idea how to fix the problem with these subs, it doesn't matter much. It's the first time I've had this problem and I've ripped a lot of subs. So it's a pretty rare problem.

And congratulations on getting married! You must be busy with all the planning and stuff. Good luck! ;)

[masken]

AfterDawn has a good mirror. Although SubRip has a project space reserved on SourceForge, we haven't begun using it yet.
Sorry but where? Link?

[ai4spam]

@masken: Google is your friend:
http://www.google.com/search?q=subrip+afterdawn
takes you to:
http://www.afterdawn.com/software/video_software/subtitle_tools/subrip.cfm

@sinistral: I'll look at it sometime, thanks for the report.

[Sinistral]

Hey ai4,

Sorry to bother you again, but I just have one quick question. Is there any way to edit the method for Post OCR Punctuation correction?

I already know about punct.dic, but not all of the method is in there. For example when you do a punctuation correction on two sentences like:
-Amen.
-Amen.
And you have an empty punct.dic, it still becomes:
- Amen.
- Amen.

I just wanna see how it works and maybe (if possible) improve it. Anyway, thanks in advance for any help and good luck with your wedding!

[LRN]

punct.dic!? I didn't know...For me, all post-ocr corrections are hadcoded (i know, i hardcoded one myself :) )

[ai4spam]

The syntax is straightforward: first line is word to be replaced, and second the word to be replaced with, and so on. So, just type in all the pairs like:
- A
-A
- B
-B
... and so on.

[masken]

ai4spam, dunno if I've asked for this before, but is there any chance of RegEx replacement support in subrip (like in Subtitle Workshop)?

[ai4spam]

I don't know what component they use, but it would indeed be fairly easy to put in, if I only had the time. My question is... if you have SubtitleWorkshop, then why not use it? Besides not supporting italics per word, everything else works nicely in it.

[awx]

Does SubtitleWorkshop open unicode .srt files? It refuses to open any of mine from SubRip.

[LRN]

Try SubtitleProcessor

[masken]

I don't know what component they useMe neither, but there's always the built-in VBscript.RegExp, and in .NET you have System.Text.RegularExpressions.

More info about RegEx and programming:
http://www.regular-expressions.info/tools.html

[Sinistral]

I completely agree with masken. It would be very nice (if you have the time ofcourse) to see Regular Expressions in SubRip.

I'm currently writing a custom script for Subtitle Workshop with Regular Expressions, but the don't have unicode support and they don't support the italics SubRip uses. Unicode support has already been asked for in the Subtitle Workshop Forums, but the develepor seems to have quit unfortunately. So it would be very nice to see it in SubRip! :)

[ai4spam]

It would be nice if they released the source ;).
As for regular expressions, it'd have to be something Delphi supports, not VB and not .NET, unless we're migrating it to Delphi .NET.
Now, assuming I'd have it done, would anyone write dictionaries/scripts for it? I haven't looked much into it, but I don't remember seeing many SubtitleWorkshop scripts floating around the web :p.

Seriously now, is there nobody woth some Delphi experience that could help? I have thought (more than once) of implementing this from scratch in C (maybe reusing code from VSRip), but I barely have time to look at the forum and fix some small bugs in the now perpetual next beta version, let alone do it all over again.

[Sinistral]

It would definately be nice if Subtitle Workshop released the source, but they won't. :mad:

I checked the Regular Expressions website and this is what they said about Delphi:
"Delphi - Delphi does not have built-in regex support. Delphi for .NET can use the .NET framework regex support. For Win32, there are several PCRE-based VCL components available."

Unfortunately, I don't know Delphi (nor any other programming language), so I don't understand what that means nor can I help you with the development of SubRip.

Fortunately, I am making a regular expression scripts for English subtitles. If I have that one finished, I might be able to delete some things out of it, so it can correct some small things for all (Latin based) languages. And if I have enough time, I might also make a Dutch one (my native language).

It will probably take some more months before they're really finished and tested, but since you probably can't implement Regular Expressions that soon, it'll be no problem.

Anyway, I'm already very happy that someone is actually listening to us and thinking about improvements, since SubRip wasn't even being developed anymore for almost 2 years. So many thanks for being here! :thanks: :thanks:

[S36d2]

Is there another mirror available with the guide for ripping hardcoded subtitles? Zuggy's site is very slow at the moment and won't show all pictures; http://zuggy.wz.cz/redir.php?co=101

[LRN]

You could use TRegExpr or T-Rex for regular expressions in Delphi.

[ai4spam]

@S36d2: I can send it to anyone who would like to host it. Send me an email (same id)@gmail.
@LRN&Sinistral: Thanks for the suggestions.

Unfortunately, SubRip has grown into a big mess, with too many features that are not really needed, and should have been implemented separately, like support for other formats. I feel very tempted to start from scratch in C, except... I don't know anything about the structure of a DVD, so I have to keep using Brain's original parser, written in Delphi. Oh, well...

[LRN]

I don't know anything about the structure of a DVD
http://sam.zoy.org/writings/dvd/subtitles/
I have to keep using Brain's original parser, written in Delphi
http://sourceforge.net/projects/subtitleripper

[masken]

I'd help out regex:ing away hearing impaired for sure :) Got loads of reference material from backing up my R1 DVD's through the years.

A humble suggestion is to move to visual studio :D

[ai4spam]

Thanks, everybody. The thing is... a GUI is still needed, and the easiest I know of is... C++ Turbo Explorer, wich is a free version of a Delphi-like GUI, with C code to drive it. The other stuff at the links seems pretty rudimentary due to the programmers being Linux-only folks, with CLI, and pretty old (unless TransCode has evolved). I don't know what to say, and I certainly don't have the time to deal with starting from scratch now :(.

[frodeste]

If you delete a character from a Character Matrix, the software will completely stop to a halt.

Is this a known bug?

[ai4spam]

What version are you using? As a workaround, use another (smaller) char matrix, it should let you delete just fine.

[Mtz]

Also my char matrix is big, and crashing SubRip when trying to delete some character. Used the latest version of SubRip 1.50b3.

enjoy,
Mtz

[frodeste]

Also my char matrix is big, and crashing SubRip when trying to delete some character. Used the latest version of SubRip 1.50b3.

enjoy,
Mtz

Same for me,

Then it seems that this is related to the size of the matrix file, and is a bug.

[ai4spam]

Sadly, it doesn't look like I'll have the time to look into it soon. It never occurred before because people were using small matrices. I guess my advice is to use smaller matrices for now.

[Teebeeke]

Well, i have the same problem when going into the matrix (no phun intended) and deleting a character.
I create a new one for every different DVD, so my matrices arent really big. Even after 8-9 characters it crashes when i want to delete one.
If i modify the character where i put a wrong input, and save it, it doesnt crash.