ANN: SubRip 1.50 b4

[ai4spam]

Well, depending on the font, that accent is grouped with the line it is closer to. You can influence it a bit by setting the "Min. inter-line height" value in the "Advanced OCR setup" window.
It is possible to do something a bit more intelligent, but I don't have the time to do it.

[ai4spam]

SubRip 1.50 Beta 3 is up.

Changelog:
- Fixed saving bug for split subtitles
- Fixed video no longer working bug
- Added a progress bar for loading char matrices
- Updated ortography dictionaries to UniCode

[Kurtnoise]

Hey,

@ai4spam : what about adding new output subtitles formats (gpac ttxt and quicktime) ? Do you have some plans to add this ? I began some stuff for gpac ttxt but from the sources of the 1.4 beta. It's ok for you or do you prefer that I start with the 1.5 branch ?

[ai4spam]

Finally someone else wants to do some developing! Yeeey!
I have no such plans, since I have no time. I'd rather work on things like accents and diacritics, and hardsubbed videos. So, knock yourself out!
Some of the changes I made between the two versions were in Unit03, where all the subtitle formats are, so you'll need to use the latest version of it. If you keep the changes localized enough and mark them clearly, I'll make sure they get into the next version.

[Sajan]

Maybe bug....
1.Open hardsubbed video
2.Play,Pause -> Run enabled
3.Run,Stop -> Prev enabled
4.Click Prev (with subtitle count =0) -> Video file viewer scroll moves to the end, Exception occurs

[Kurtnoise]

I have no such plans, since I have no time. I'd rather work on things like accents and diacritics, and hardsubbed videos. So, knock yourself out!
ok...just a suggestion : have a look on the ï french special character...detection sucks. But this is not related to your changes. We have this problem since the beginning.


@Sajan: bug confirmed...:)

[ai4spam]

@Sajan: I noticed, just forgot to take care of it.
@KurtNoise13: Until then use "take with next", that's what it's there for.

[ai4spam]

A request: would someone be willing to mirror the hardsubbed avi guide? It loads very slowly, and probably starves the bandwidth for the software downloads. Drop me an email (it's in the SubRip manual).

Thanks,

[johner23]

@ai4spam:

I am ripping some specifical movies here and find some specifical Monetary Symbols to rip. Can you add some of them to put into Subrip as standard?

See some monetary symbols that you could put into Subrip, like "£" - Pound sterling ( I had to find these in some sites on internet and copy / paste them to put in my subs before save my final srt) or euro, for example.

---> http://en.wikipedia.org/wiki/Talk:Pound_sterling

---> http://en.wikipedia.org/wiki/Pound_Sterling

---> http://www.typography.com/catalog/hoeflertext/altchar.html

---> http://www.paratype.com/help/term/terms.asp?code=226

---> http://www.microsoft.com/typography/developers/fdsspec/monetary.aspx

---> http://www.microsoft.com/typography/developers/fdsspec/monetary.htm

And there is some other symbols, like in Math or measure/conversion numbers and scales, can you just add or find a way to add in some cases, for special OCR cases?

Of course, they are very rare to appear on subtitling scene, but it happens sometimes. :)

Thanks.

devil (johner)

[ai4spam]

All you have to do is add tem in the Punctuation section of CharMaps.ini. Edit it with NotePad, and paste them in there, then they'll show up in the Punct. row of the table.
Beware, though, they are UniCode chars, so you'll probably need to save the subs as UniCode. That's why, although there is a character for the "note" used to show songs, I use "#" instead.
As for Math and the like, just make a new "language" called "Math" in CharMaps.ini, and type in all the characters after "Small=", separated by "*".
If you do this, please send me the resulting CharMaps.ini file to share with everybody.

[ai4spam]

@KurtNoise: If you're adding new output formats, don't forget to change everywhere: Unit3 for saving, Unit13 for options and converting, UConfig for loading/saving settings, etc.

[johner23]

Ok, then... :)

And what about musical symbols like that above?

---> http://www.dolmetsch.com/musicalsymbols.htm

so you'll probably need to save the subs as UniCode. That's why, although there is a character for the "note" used to show songs, I use "#" instead.

Only using Unicode I can reproduce music symbols and similar ones properly in my subs? No other way?

I am asking that because I always use "srt" format to save my subs. These sub format has some limitations about that issues, then? :(

Ah, just another subtitling question...

I saw one videoclip from an rapper singing in jamaican patois style. For those who listen Snow ( performing the song Informer), knows what I am talking about.

The guy sang so fast that the TV editors must have to create a "different type of sub" to show correctly the letters on screen. They create a rolling bar that, literally, rolling on the video showing the lyrics from the music.

It was very, very fast. We have some difficult to follow with our eyes too.

Do you know if exist any sub format that can handle or create some kind of effect on subtitles? Or I wil need to use some professional Video Editor like Adobe Premiere, Sony Vegas or similar to create such effect and hardsub the subs to the video later?

Thanks.

devil (johner)

[ai4spam]

Well, VSfilter UniCode supports UniCode .srt files, so no problem there. It'll be a while until it is universally accepted though: basically all the utilities will need to learn how to deal with the new format. As for "reproducing", well, the result is like using Word as your music editor. Not really what you want, is it? I think bitmapped hardsubs are the best way, unless osmeone comes up with some format that is universally accepted.

You can simulate rolling subs by displaying different parts at any moment. I think SubtitleWorkshop lets you "decompose" a line into parts (the Effects menu). However, you need breaks in between, so that lines can be decomposed properly. I know of no utility that does this right now. Maybe you should write one ;). Here's an example for "The quick brown fox jumps over the lazy dog":

The quick brown
he quick brown
e quick brown f
quick brown fo
quick brown fox
uick brown fox
ick brown fox j
ck brown fox ju
k brown fox jum
brown fox jump

...and so on
Timed properly, this does what you want. There are sub formats for karaoke (not sure which) that have 2 text colors and show text being colored while the song goes on, but this thread is not the right place to about them. I'm sure there are some knowledgeable people in the parent topic group.

[johner23]

Thanks for helping me, ai4spam. :thanks:

Maybe TitleDeko Pro can create some scrolling/rolling "video" text bar, that can subtitle very fast speeches during one video clip, for example.

---> http://www.pinnaclesys.com/PublicSite/us/Products/Consumer+Products/Advanced+Video/Plug-In+Edition/TitleDeko+Pro.htm

The only problem is that HOW to time that, because I think "srt" doesn't support this kind of features. I dunno. I will try here some experiments and looking some other advices with some people who works in Karaoke/animes fields and video editing too.

Well... concerning about SubRip, I have some videos in my home that have a very hardsub white pale subtitles to "extract".

Any special set to do it in Subrip? I ask that because the golden/yellow ones are pretty much easier to be OCR'ed than the white ones. Specially if the scenes are very clear too.

One more thing: there is a way to "accelerate" ( get faster) the process of ripping/reading harcoded subs in Subrip? Ripping DVD's is much faster, a question of few minutes in some cases.

I know that reading/OCR in AVI's and similar ones are much more complex than one from separate streams like in dvd's. But, maybe, you can find out some way to improve the reading speed or something like that.

PS: for your experience, one movie that has 2 hours lenght takes ( generally, of course, because it depends case by case) how many minutes/hours to OCR a hardsub from it?

Thanks for your program. Very, very useful for me. The last week I use it more than 10 times !!! ;)

devil (johner)

[ai4spam]

Of course .srt supports timing. As per my example above, one line becomes 10 lines, each with 1/10 of the initial duration.

Unfortunately, if the subtitles don't have an outline, any program will have a hard time extracting them. Try AviSubDetector too, if you can figure out the options - it lets you do a lot more tweaking, and its detection methods may work better for some particular cases.

Accelerating is possible, but I don't have much time to do it these days. I have a few ideas to make it work faster. The bottleneck isn't in the processing, it's more in the fact that chars with minimal differences are not recognized, and SubRip asks you to type them in again. You can try to lower the OCR sensitivity, then run the whole thing through a spellchecker at the end. I can usually rip subs from a 2-hour movie in 1 hour, even faster depending on the quality of the subs.

[ukendt]

For several reasons, this thread is going to be unsticked. We'll start a new, sticky one

[ai4spam]

We'll start a new, sticky one
...when another beta is up. No telling how soon that will be, though ;).

[ukendt]

:thanks: (for the specification although I wasn't going to tell anyone anything:D

[ai4spam]

I didn't want anyone refraining from posting bug reports because they had to wait for the new thread ;).

[Paulcat]

I tried ripping subs from an old dutch film, and what would happen was that the letter in question (I started with a blank character matrix) would appear with a red box around it, but the rest of the subtitle bitmap would not appear in the main window.

I had some difficulty distinguishing a capital I from a small L without the rest of the words visible and as a result had a lot of editing to do. Anyone else have the same problem? BTW the 'display subtitles' box was checked...

Paul

[ai4spam]

It's just the color palette in that particular subtitle. Can you send it to me (look up my email in the manual), maybe I can find a way to deal with subs like it?

[SpellCaster]

Hi everybody, good day to all!
I've checked out latest SubRip version just yesterday (previously I was using ancient 0.9 ver.). There's much improvements since that version, but I've found some problems with new one.
1. Window labels have VERY small font! I had to get much closer to monitor to read them :(. Especially it applies to Subtitle combobox.
2. When ripping, main window doesn't keep its size/position. At some moment (i believe when the one VOB is finished and a new one started) window returns to default placement, whatever placement/size it had before.
I'm now downloading source, maybe i'll tell where this bug located.
3. The red blinking of "Enter Char" window is REALLY annoying! This window has a lot of buttons, fields and stuff, and i was just studying them but red blink almost drove me mad >:-[. It would be nice to have an option allowing to turn the blinking off.
4. I didn't found an option to set inter-character distance. In Russian, letters are often flowing together and SubRip considers that 2 or even 3 chars are one.
5. I've tried to fill the matrix using some font, but nothing was filled, although subtitle font was like Verdana. Maybe i did it wrongly?
6. (trifle or maybe my OS bug). Scrolling in window where text is stored does by strange way. When scrolling (with mouse wheel) down, it seems like scrolling up and then down 8-|

Best regards, SpellCaster.

[bourtzovlakas]

1)You can raise the font size...
3)You can disable it from Options-->Global options-->Wake me up:untick
4)Options-->Advance OCR setup-->"Space width setup" and "character setup"

[ai4spam]

2) I usually do .idx/.sub files, so I never noticed this. Let me know if you find the bug.
5) Is currently being improved (I got a new, automatic version in the making). But, in a nutshell, you need to select some UniCode ranges that will appear in the "characters to fill" box.
6) You may have peoblems with your mouse driver settings as well (there's a setting telling it how many lines to scroll). Also, the "V" checkbox tells SubRip to always scroll to the last line when adding a subtitle, and that may conflict with your manual scrolling.

[SpellCaster]

Thank you for answers!
1. You mean, set font size in Delphi and then recompile?
2. I will take a look at the code
3,4. Thanx, i got it

[ai4spam]

1) There used to be an option that let you set the interface font and size. I'll think about reintroducing it in the next version. Meanwhile, you can tell Windows (Desktop->Properties->Appearance) to scale up your fonts. Or... get a pair of glasses ;).

[ToS_Maverick]

hello!

i've been using subrip for quite a while and it's great!

there is one major improvement that i'd like:
is it possible to select which subtitles should be ripped and where they are saved? it's a pita-work to select each subtitle track, wait for it to be finished, save it, clear the text, reset the time-count and start to rip again.

it would be nice to set everything up and let the pc do the work.
i know that i still have to enter the correct characters if SubRip doesn't know them, but after 4 or 5 episodes almost all characters are known.

[ai4spam]

SubRip supports comand line interface. Just read CLI.txt for details.

[ToS_Maverick]

yeah, i tried it,but i need .srt files!

what i've seen so far, subrip doesn't work with CLI + OCR?

[ai4spam]

Indeed, since SubRip assumes that the need to type in characters defeats the whole purpose of the command line interface (no interaction). As more featurea are added (better OCR, planned in the next version), text output for CLI may be implemented, but I don't have any time for it in the near future, so I can't tell you if/when it will be done.

[ToS_Maverick]

thank you very much for the information!

with your character-matrix, text output shouldn't be a problem ;)

[jmk]

ai4spam: would you be so kind and put your char matrix file online again? its hosting period seems to have expired...

thx for the great program!

jmk

[CoNS]

Try this:

There's a mirror (kindly provided by FoxyShadis): http://foxyshadis.slightlydark.com/random/CharMatrix.rar

ai4spam, you may want to correct the links for the charmatrix at the subrip homepage etc.

[jmk]

sorry,
should have probably read through the whole thread... i came from the subrip homepage.

thanks for the link / mirror!

jmk

[foxyshadis]

Of course, if you ever need me to update it, just PM me! It's about six months old, I have no idea if it's changed at all in that time...

[promitheas]

" I didn't found an option to set inter-character distance. In Russian, letters are often flowing together and SubRip considers that 2 or even 3 chars are one."
I agree with Spellcaster, the same happens in Greek language.

[ai4spam]

@foxyshadis: Thanks a lot for all your help. Not much ahs changed in the matrix since the last update. It would be great if you could mirror other stuff as well, such as the hardsubbed video ripping guide.

@promitheas: Space width setup in Advanced OCR options will do that for you. Bear in mind that if the letters are "glued" together, there's no straightforward way to separate them.

[ericf]

Hello. Sorry to say this but I've not been able to make this work with any of my files. The OCR feature simply doesn't work for me. I can't get the selections to work for more than one subtitle ( it seems the subs move up and down all the time), I can't make the program fill in the background in any color but red (it's never black and white), fatten text doesn't take care of fonts with thin bends (like Times New Roman). I got the program to find and react to the first appearing subtitle at some point with one file but when I paused adjusted the parameters to make it read the subs better (broken subs) and started over the program didn't detect any subs at all.

And then, if I did get some of these right, the process is very slow.

Am I doing something wrong here or is this what people have to cope with all the time? I know that subrip doesn't know back from front when it comes to characters with broken parts. And I have no way of making my subs appear fully even with fatten text.

Another thing. If I wanted to exit from the ripping window and go back to the startwindow of subrip, how do I do it? I can't shut down the window since there is no close button and no X.

Thanks.

ericf

[ai4spam]

If the subs move, don't check "Fill sides" or "Draw lines on top". The background may be red because you tell SubRip to use the outline color, which happens to be close to whatever else is in the image. Try adjusting the color tolerances. And look in the status bar at the bottom, it tells you what to do. You have to click a second time to define the outline color. As for the processing being slow... it is the best it can do at the moment, and I don't have time to improve it in the near future, sorry. You may want to try AviSubDetector as well, it may work better for a particular movie.
And... to exit, press Stop, then Close. You can also use the Pause button to change some settings, Resume will go back to the first frame of the current subtitle.

[ericf]

Thanks for the tips. I will try to work it out.
I have used AVISubDetector for about a year and I'm quite happy with it's detecting abilities. The OCR is quicker but when fansubbers use different outline colors depending on viewpoint and sometimes different fonts it's quite difficult to OCR no matter what program you use.

Thanks again for putting in effort making this program.

By the way, how do I make an Avisynth script for an mpg file (or any avi file as well) and open it in AVISubDetector? The problem is that the video is shown upside down and the flip command doesn't work for the program.

Do you know? I've tried to get an answer from the creator of that program but failed.

ericf

[ai4spam]

I'm not that familiar with it myself. SubRip just uses 2 commands, depending on whether the movie is an .avi or another type. If it's anything but .avi, it'll open it using AviSynth DirectShow interface. Look in the SubRip directory for the file named temp.avs, that's what SubRip writes commands into and loads.

[Kurtnoise]

@ai4spam ::

Hi...;)

I've got a question for development : do you know a way to convert bmp files ? I mean to change the colors (I've seen graphics unit but...) and especially to have the alpha channel ?

[edit]: hihi, I found something (http://members.chello.be/ws36637/transparent3.html)...

[ai4spam]

There should be a ton of components on torry.net that support transparency and conversions.

[Zerryk]

I don't know if I am doing something wrong or there is an error... It seems that it can not open MicroDVD .sub format. Only .srt seems to be working, I didnt test it for other formats.

Steps to reproduce:
- make a clean install of 1.50b3 into a new directory
- run subrip.exe
- show subtitles text window
- create simple test.sub file like that:
{1}{10}aaa
{11}{20}bbb

or take any other correct MDVD .sub subtitles.
- open it in the text window, via drag'n'drop or via File-Open
- an error dialog appears:
Warning
Error in file at Subtitle Nr.1
Ignore and Continue?
[Cancel] [Ignore] [Yes to all]
- None of the buttons helps.

It also does the same error in 1.40 version. In 1.30, there is no such bug, all works perfectly.

--------------------
And some idea for opening and saving the text files: When subs contains some non-7bit characters, the asking for the charset when opening, also when saving, and the warning dialog "Subtitle contains non-standard characters, save as Unicode?" is a bit annoying. In most (nearly all) cases, I'm working with one language only so I dont need always to watch and confirm these settings. I think it would be beter to:
- make these confirmations optional
- duplicate the charset setup somewhere in program options
- remember it for next program run

[prubin]

Hello, I have been a very happy SubRip user for some time, and while the OCR works great with most .idx or .sub files, every once in a while I come across one that has a few lines in which SubRip seems to want to pick apart each letter in the whole line. For instance it will take a simple word, like 'problem', and break each of the seven letters into two or three pieces. Typically this problem continues until the line is complete, and then the next lines it recognizes fine, until the next problem line.

Needless to say, the problem lines are recognized as gibberish, and I have to go back and extract the sub pictures as .bmp's to manually do the conversion.

Any ideas?

[ai4spam]

@Zerrik: I'll look into it.
@prubin: It probably has to do with the wrong colors being assigned (heuristics sometimes fail). Try to assign them manually ("Change colors" button).

[prubin]

Thank you for your reply. I'm working with one now that has white characters on a black ground. I don't think changing colors would do anything. Let me know what you think.

@Zerrik: I'll look into it.
@prubin: It probably has to do with the wrong colors being assigned (heuristics sometimes fail). Try to assign them manually ("Change colors" button).

[ai4spam]

Well, try changing the colors anyway, see what it does, if anything. The colors you actually see may have very little meaning (for example, the outline color may look the same as the text color). Oh, and use the Extend Left and Extend Right buttons to bring together "disjoint" characters as a last resort.

[prubin]

Thank you. I did as the 'Change Colors' box suggested: "leave background and all outlines unchecked. This setting will verride the built-in autodetection." and amazingly, it worked! No split characters, and no garbling. Thanks for the tip!

I don't see the 'Extend Left' and 'Extend Right' buttons; where are they?

Well, try changing the colors anyway, see what it does, if anything. The colors you actually see may have very little meaning (for example, the outline color may look the same as the text color). Oh, and use the Extend Left and Extend Right buttons to bring together "disjoint" characters as a last resort.

[zyrill]

I OCRed some vobs and when I select to save the Subtitles, the resulting file is empty except for a lot of empty lines. :s Tried selecting different Codepages, to no avail.