Subtitle Edit 3.5.17

[FLX90]

Currently I'm using Tesseract 4.00 with engine mode Tesseract + LSTM.

Unfortunately I have to do massive post work.
'I' is sometimes a 'L' ,'l' or '!'.
Music symbol isn't recognized right (it's a P or D).
And I don't know what I have overlooked.

Is there anything I could improve in my settings?

EDIT:
I'm using binary image compare now and creating new database for every BluRay.
Works.

[johner23]

Hello, dear all.

I got the newest version from Subtitle Edit and try to convert a subtitle format to other.

The original subtitle is *.vtt fomat. The output is to be *.srt format.

But after the convertion the subtitle in srt show some strange errors.

Eg:

00:01:04,562 --> 00:01:08,908
{\an3}>> O que mais lhe
interessa na história?

The symbols

{\an3}

and

>>

are not correct.

Do you know how to fix that?

I can make manual correction after the convertion to srt, but I would prefer that the program could do it at the first attempt, without any "errors" or strange "symbols" after the task.

Thanks for your time.

Best regards.

[Hendi]

Hi

If I convert from sub to srt on batch-mode (/convert) the resulting srt file contains all timeframes but only the first timeframe contains a text. If I do this with the GUI, the whole text is converted and all timeframes contains text.

What do I wrong?

Thanks

Hendi

[Nikse555]

@Hendi: That's a bug, sorry - should be fixed in latest beta: https://github.com/SubtitleEdit/subt...leEditBeta.zip
Next final version should be out soon...

[asarian]

Originally Posted by Nikse555

Subtitle Edit 3.5 is now out

(...)

but SE can also [B]import and ocr vobsub and blu-ray image based subtitles (even from matroska/mp4 files), and DVB sub from .ts files

Subtitle Edit (the latest) actually immediately crashes for me when I import a .sup file ("Object reference not set to an instance of an object." and many other fatal errors).

Btw, why OCR-ing?! That's so 1985!

[Nikse555]

Originally Posted by asarian

Subtitle Edit (the latest) actually immediately crashes for me when I import a .sup file ("Object reference not set to an instance of an object." and many other fatal errors).

Could you try latest beta?
https://github.com/SubtitleEdit/subt...leEditBeta.zip

SE 3.5.9 should be out soon...

[dngnt]

Originally Posted by Nikse555

SE 3.5.9 should be out soon...

I'm using SE to convert from bitmap Chinese/Japanese/Korean subs (sup, idx/sub, DVDsub).
There is a possibility to save one sub picture at a time (with right click) while importing , but could you add a function of saving all the sub pictures during OCR?

For example, just a new option in the "export" (during OCR): export all the pictures as *.png? (just the images, no xml because it would be more complicated with "dirty" idx/subs (i.e with errors).)

It's because after finishing the OCR and saving the OCRed srt, I have to correct the (many) mistakes "offline" , but there is no way to compare the OCRed text to the original pictures.

Thanks for your great work!

[Nikse555]

@dngnt: In the OCR window you can right click in the list view... and choose "Save all images with HTML index..." or "Export -> BDN xml/png"

[Nikse555]

And SE 3.5.9 is out: https://github.com/SubtitleEdit/subtitleedit/releases

Some of the changes (mostly related to OCR) listed below:

* NEW:
* Bookmarks - thx OmrSi/marb99
* Image export - option to have single lines top justified - thx joedmartin
* IMPROVED:
* Improve Binary OCR of comma / apostrophe - thx Tuukka
* Improve quote/italic detection in binary OCR - thx Miggu
* Add context menu to OCR spell check
* Binary OCR auto detect best DB - thx Mr. Rage
* FIXED:
* Fix missing/bad html tags after "Auto br" - thx iromafia111
* Fix crash in OCR window when closing - thx spetragl
* Fix OCR in batch convert - thx danstraughn
* Fix crash parsing empty word in OCR via Tesseract - thx Barry

[varekai]

@ Nikse555

Thanks for the update, much appreciated!
Looking in the portable folders I see Tesseract302.
Is that version also included in SubtitleEdit-3.5.9-Setup.exe?

Kind regards

[Nikse555]

@varekai: Yes, Tesseract302 is included in the installer version as well.
Tesseract 4 is available as an in-program download, but note that T4 does not support italic detection and is a lot slower, but T4 is available in more languages and may work better for small unclear fonts.

[varekai]

@ Nikse555

Great! I use italics and only swedish and english languages so Tesseract302 works perfectly for my needs! Thanks!

[dngnt]

Originally Posted by Nikse555

@dngnt: In the OCR window you can right click in the list view... and choose "Save all images with HTML index..." or "Export -> BDN xml/png"

Thanks for your indications!
Subtitle Edit is the best for its import/export versatility for srt, and with these 2 export abilities for sup,idx/sub and DVDsub, it's truly the Swiss knife for subtitling!

[johner23]

Hi, dear all.

@Nikse555

Hi, Nikse.

When I tested some files *.vtt I still get some errors. Different ones, I mean. Please, can you check it?

---> https://github.com/SubtitleEdit/subt...it/issues/3290

---> https://github.com/SubtitleEdit/subt....-.SAMPLES.zip

I can fix it manually using text editor, but I think it will be better if your program could do it automatically for us.

Thanks for your time.

Best regards.

[Nikse555]

@johner23: thx for the files - latest beta handles this better: https://github.com/SubtitleEdit/subt...leEditBeta.zip
SE adds/keeps the alignment though, so that why you'll see lines starting with e.g. "[\an8}". If you want to remove them, just select all lines in the list view (ctrl+a), right click, choose "Remove formatting -> Remove all formattings / Remove alignment".

[locotus]

Is there any way of sort lines by beginning capital or smal letters or
by lines beginning with small letters fallowing lines ending
with comma, space or other punctuation mark except period?

Thanks.

[sneaker_ger]

Batch converting vobsub in mkv via cli /convert (lots of "Tesseract returned with code 1" messages, timestamps are there but lines are empty) as well as GUI Tools->Batch (lots of "Tesseract returned with code 1" messages, almost empty files) convert isn't working for me. I would also be nice to have a stream selector for mkv/mp4 input on the cli.

[Nikse555]

@locotus: sorry, no - what exactly do you want to do?

@sneaker_ger: that should be fixed in last beta I hope - could you verify?
https://github.com/SubtitleEdit/subt...leEditBeta.zip (track-number parameter also added to cmd line)

[locotus]

@locotus: sorry, no - what exactly do you want to do?

My main purpuse is to join dialogs lines that are
divided but with duration time and number of characters
below maximun range for both.

Actually I'm trying to do that sorting first for duration,
looking for very short lines with characters number greater
than 15. But can't do that with lines with longer time and
characters.

Thanks anyhow.

[sneaker_ger]

Thx. I think there's a small typo: /? says "/trac-number:<track number>" (missing "k")

It still showed "Tesseract returned with code 1" over CLI but I figured out why: in the GUI OCR method was set to "Binary image compare". I changed it to "Tesseract 3.02" and the messages disappeared, the text appeared. I didn't expect the GUI setting to influence the CLI. Also it's not obvious why I would get Tesseract error messages if Tesseract isn't selected.

If you drag&drop mkv files into the GUI batch converter the GUI hangs for a long time, btw.