Subtitle Edit

[tormento]

@tormento: thx for the info
Thanks for fast fix. Wil try for some days before reporting.

P.S: I suggest you to add text removal before ":" also for mixed case option, such as "BiLBO:" and so on.

[varekai]

OT: uBlock Origin is the cure.
Looked into it and the uBlock Origin reviews aren't hot...
Why you won't use mediafire is beyond me.

Regards

[rco133]

Hi.

Got a question about the spell checking.

After OCR I often end up with a lot of lines with "I" (capital I) actually being a l (low capital l) or L (capital L).

Words like

l'm should be I'm
L'm should be I'm
ls should be Is
lt's should be It's

When running a spell check, it happilly accepts all those Words, eventhough none of them I think is even valid English Words.

Also "Bob." became "Bob-ll

Which might be valid, but should still be pointed out by the spell checker?

I am not even sure if spell checking is an internal part of Subtitle Edit. If not I guess there is nothing to do about it.

I just think that the spell checker accepts a lot of Words, that are not even valid English words.

Maybe it is me not using it correctly. If so please let me know how to make it catch all these words.

If not, can something maybe be changed to make it catch all these words?

Thanks in advance.

rco133

[Nikse555]

@rco133: SE uses Hunspell... I would prefer not to add extra code with special OCR related hardcoded rules in the spell check.


ATM you can either:

1) Use Tools -> Fix common errors - Fix common OCR errors

or

2) Make your own rules via Edit -> Multiple replace

[Ghitulescu]

Probably half of this thread is concerned with partial solutions to a problem that should never existed in the first place.

Why do these spelling errors arrive and wherefrom? Can't the user correctly write in his own language?

Of course I know the answer. The errors comes from OCR. Geez, we do not have enough problems, we are bored, let's convert from the original subtitles to another one, to save only a couple of kB, to occupy our minds then with issues like code page (for anyone not English), L or I, # or ♫, colours, timings, positions and of course the compatibility of the result with the existing players. Players that otherwise will play the original file just perfect, and 83-years-old-age compliant.

Some people want the dash before the first line of dialogue, others request its deletion. Go figure.

Issues and problems that concern a secondary if not even non-related task of the software, a bonus functionality if you like.

Meanwhile issues that are part of the core, like loosing the position on-screen of the original subtitles (these are centered after processing) do not have time to get solved, because someone has a spelling issue with tembo-tembe subdialect and semicolons :) .

[DMD]

Good morning.
I have a problem with DirectShow (W10 @ 64bit) since a few days has stopped working and in the warning of Subtitle Edit I get a black screen but the audio is present.
I am forced to set up VLC as a video player in the Options> Preferences.
I can not understand what happened, :( can someone give me some advice?

Thank you

[Nikse555]

@DMD: It's probably because you're missing a decoder for the video - LAV filters contains decoders for the most common used formats - get it here: https://github.com/Nevcairiel/LAVFilters/releases (use the installer)
(if you're sync'ing or creating subtitles I recommend mpv as VLC does not have precise seeking)

[DMD]

Thank you for your help.
I installed LAV filter 0.72, but I still have the same problem.

https://s20.postimg.cc/hyf3lhbzh/Screenshot_001.png

https://s20.postimg.cc/a5oftidq5/Screenshot_002.png

[Nikse555]

What version of LAV filters did you install? SE probably runs as a 64-bit program on your computer, so it's important you did not install only the 32-bit version. (I think the installer installs both versions per default).
MediaInfo can tell you what video codec you're having trouble with.

Perhaps someone else has some good ideas?

[LowDead]

No good ideas, but I actually have the same problem since a while back.. Haven't gotten around troubleshooting yet though.

//LD

[DMD]

FINALLY I RESOLVED !! :):)
After endless tests, I discovered that the problem was due to the latest nVidia driver release and not a codec problem.
The latest drivers are version 398.82 for the GeForce GTX 960 video card, I tried to install version 390.65, with this version I solved all the problems.

Thank you all for the availability

[varekai]

The latest drivers are version 398.82 for the GeForce GTX 960 video card,
I tried to install version 390.65, with this version I solved all the problems.
Not the drivers fault, somethings wrong with your setup.
Works great for me with 398.82.
BTW, latest drivers are 399.07.

Update:
Works perfectly with 399.07.

Regards

[DMD]

I created the current image of the pc through Acronis, I have formatted and reinstalled only the 399.07 Driver the video with directShow turns black.
With the regressive driver test, I checked the regular operation with version 391.35.
I think the problem could be the GTX960 video card that has problems with newer drivers.

Regards

[Zetti]

Thanks for new release:
https://github.com/SubtitleEdit/subtitleedit/releases/tag/3.5.7

[jpsdr]

It never happened to me before, i've until now always used the portable version, but with this last version, i can't start OCR, i have popup and error message telling me that tesseract can't be started, to check everything is properly installed.
Everything is properly installed (redistribuable 2017, Framework .NET 4.7.1). Under Windows 7 x86.
Rolled back to 3.5.6, everything worked fine again.

Edit :
Could this be related ?

Top Issues Fixed in 15.8.3

These are the customer-reported issues addressed in 15.8.3:

Visual Studio 2017 version 15.8.2 contained a pre-release build of .NET Core SDK 2.1.401 that is incompatible with Visual Studio. We have corrected this issue with Visual Studio 2017 version 15.8.3.

[varekai]

Many thanks for the new release! SubTitle Edit is one-of-a-kind!

Installed 3.5.7 and kept old settings.
Started a new project today and noticed some new stuff in GUI.
Didn't change anything.
Ran the new project and there was a noticeable sluggish performance.
Usually it just flows through but now it stumbled and almost froze for a second before moving on.
Tested with old projects and the sluggish behavior is there.

So I rolled back to 3.5.6 and all is good, both the new project and old ones and the sluggish behavior is gone.
I have no clue on what this is about or what I can do.

Best regards

[Nikse555]

@jpsdr: The Tesseract version used by SE is now "Tesseract 4.0 beta 3" which might need a newer "Visual C++ Redistributable for Visual Studio" - could you try to install this: https://www.microsoft.com/en-us/download/details.aspx?id=48145
Does that help?
The reason for upgrading from Tesseract 3.02 to 4 is mainly due to support for more languages... Tesseract 4 is a lot slower and it's also the reason for the SE increase size (4 extra mb)!

@varekai:The sluggish behavior is when doing what exactly? Did you change video player?

[varekai]

Sorry for bad explanation...
I have first window open (List view) and then File --> Import/OCR Blu-ray (.sup) subtitle file...
No video player active in this scenario, when I use video I have mpv selected.
Havn't tried that yet in new version but it worked perfect in 3.5.6 last time I used it.

[jpsdr]

Euh... Your link is for a VS2015 redistribuable. You can't install redistribuable 2015 when you've installed the redistribuable 2017. I have installed the 14.15.26706 version of 2017 redistribuable, i have Framework.NET 4.7.2, tesseract still don't want to start.
Not a big deal, for now, i'll stay with 3.5.6.

[jpsdr]

More information : I have a big popup with a lot of informations, but the first line, translated said : This is not a valid application for this operating system.
As i've notify in post #716, i am under Windows 7 x86.
Meaning your tesseract version either is not compatible Windows 7, or is not for x86 OS. Have you tested either under Windows 7 and/or under 32 bits OS ?
After, if for now versions are not Windows 7 compatible or are not 32 bits possible anymore, Ok, but at least tell it.

[Nikse555]

@jpsdr: thx, it seems that I've included the x64-bit version instead of the x86 version of Tesseract... I'll upload a new beta later today.

[varekai]

@Nikse555

The reason for upgrading from Tesseract 3.02 to 4 is mainly due to support for more languages...
Tesseract 4 is a lot slower and it's also the reason for the SE increase size (4 extra mb)!
Is this the reason I'm seeing a "sluggish" performance?
Are there any settings in new version I could use have the flow as in 3.5.6?

Best regards

[Nikse555]

@jpsdr: Could you test latest beta: https://github.com/SubtitleEdit/subtitleedit/releases/download/3.5.7/SubtitleEditBeta.zip
How is that?

@varekai: You could try the above beta.

[varekai]

@varekai: You could try the above beta.

Hello!

Just tried the beta and experienced the same sluggishness.
Imported a Blu-ray sup.
Also I noticed a lot of weird linebrakes that is not in 3.5.6

This is a screenshot from 3.5.7
https://i.imgur.com/Q4sY5Ek.jpg
This is from 3.5.6
https://i.imgur.com/yRJ5mCK.jpg

Best regards

[DMD]

Good morning.
I'm sorry, for the trivial question, but I can not select MPV player.
I downloaded the file from the site https://mpv.io/installation/

https://s20.postimg.cc/y7vpoo0j1/Screenshot_001.png

The version is in portable format, I also tried to press "Download mpv lib" but I can not select "mpv".
How should I proceed?
Thank you

[Nikse555]

@varekai: Can you upload the source image? (you can right click on it in SE and save it)

@DMD: You need to click "Download" again in the popup window.
You can also download the "mpv-1.dll" from the dev package - be sure to take the release version in the correct x64 or x86 version. Atm latest dev package is here: https://mpv.srsfckn.biz/mpv-dev-20180731.7z

[DMD]

@DMD: You need to click "Download" again in the popup window.
You can also download the "mpv-1.dll" from the dev package - be sure to take the release version in the correct x64 or x86 version. Atm latest dev package is here: https://mpv.srsfckn.biz/mpv-dev-20180731.7z

Thank you very much, I finally solved! :):)

[jpsdr]

Ok, i've tested, it works now, but my feedback on a sup file i've tested :
- Too slow.
- Too much more errors. Tesserac 3 produces realy significant better results.
For me, 3.5.7 is only negative. So, question :
- Is it possible to still use the previous tesserac version if i replace the files in the tesserac4 directory ?
- Is it possible to have both versions of tesserac, with two directory "tesserac" and "tesserac4", and a global setting in the "configuration/tools" which select the tesserac version used ?

[varekai]

@varekai: Can you upload the source image? (you can right click on it in SE and save it)
This is from 3.5.7
https://i.imgur.com/ZMtUhOK.png

Best regards

[Nikse555]

@jpsdr: OK, I've tried to add Tesseract 3.02 back in also... how does this work: https://github.com/SubtitleEdit/subtitleedit/releases/download/3.5.7/SubtitleEditBeta.zip ?

[jpsdr]

When i select 3.02 in this beta version, i have more errors than with the 3.02 from the 3.5.6.
I've replaced the whole 3.02 (downloaded) directory by the directory from the 3.5.6 (for testing, just in case), but no change in behavior.

[RyaNJ]

Hi there.

I've got a quick question for you all. I've run into an issue with the BD subtitle import (.sup). When they are subtitles at the top and bottom of an image it doesn't seem to correctly identify this and either ignores one or puts them both into a single line at the bottom, leading to knock on issues. It could be that I'm doing something wrong since I am new to this tool. Does anyone have any experience with this? Is there a way around this that doesn't require manually editing these every time?

Thank you kindly!

[Ghitulescu]

I've run into an issue with the BD subtitle import (.sup). When they are subtitles at the top and bottom of an image it doesn't seem to correctly identify this and either ignores one or puts them both into a single line at the bottom, leading to knock on issues. It could be that I'm doing something wrong since I am new to this tool.

It appears that the spatial position of the subtitles is lost during importing.

[RyaNJ]

It appears that the spatial position of the subtitles is lost during importing.

That's more than slightly annoying. If there isn't any way around that then does anyone have a suggestion as to a tool that can actually do it?

This is by far one of the best editing tools that I've used so hopefully there is some way to make this work though :)

[Ghitulescu]

If there is one I haven't found it.

[Nikse555]

@jspdr: Yes, SE 3.5.7 should probably not have included Tesseract 4 by default... I've uploaded a new beta here: https://github.com/SubtitleEdit/subtitleedit/releases/download/3.5.7/SubtitleEditBeta.zip
Do you still have more errors with 3.02 in the new beta than with the 3.02 from the 3.5.6? If yes, should you share your sub?
Other changes: If you right-click in the list view you can also check the new "image pre-procissing" + Tesseract should also be faster if you have a nice CPU with many cores.

@varekai: Yes, Tesseract 4 is not really super... you can change the engine mode to "Neural Nets LSTM" which is the new engine - but it does not have support for italics and it also seems to have problems with line breaking.

@RyaNJ/Ghitulescu: Sorry, SE does not support positioning of subtitles much. Do you need only like "align top, left" or the actual position like "100, 300" ?
Also, you could try https://www.videohelp.com/software/SubExtractor

[RyaNJ]

@RyaNJ/Ghitulescu: Sorry, SE does not support positioning of subtitles much. Do you need only like "align top, left" or the actual position like "100, 300" ?
Also, you could try https://www.videohelp.com/software/SubExtractor

Thanks for the reply :) Nah, I don't need the exact position - I literally just need the subs at the top to stay at the top and the ones at the bottom to stay at the bottom. I could do it by hand but that would take days.

I'll give that a try, thanks!

[varekai]

@varekai: Yes, Tesseract 4 is not really super...
you can change the engine mode to "Neural Nets LSTM" which is the new engine
- but it does not have support for italics and it also seems to have problems with line breaking.

Hello Nikse!
Thanks for the follow up and clarification, appreciate that!
For now I will stay with 3.5.6 which works very well for my needs.

Thanks and best regards

[RyaNJ]

Thanks for the reply :) Nah, I don't need the exact position - I literally just need the subs at the top to stay at the top and the ones at the bottom to stay at the bottom. I could do it by hand but that would take days.

I'll give that a try, thanks!

I just played around with it a bit and while it works... well... let's say the resulting mess is worse than having to fix a few lines with simultaneous top/bottom subtitles here and there. It's OCR isn't the best (given it's age that isn't a surprise really) and the resulting mess is difficult to clean up.

I think I'll just have to give up on the idea and just leave the subtitles alone. I don't have a hour to spend cleaning up each one of these videos as that would take weeks. I just don't have that sort of time. If Subtitle Edit can ever handle detection of top/bottom subtitle elements then I will revisit in the future.

Thanks for the suggestion none the less :)

[Matt Kirby]

Is there a way to get "Auto break" splits the text in 3 lines when there are to many chars?

in options I set "max number of lines" to "3". Now I want that "auto break" uses this option, but it allways splits the text into 2 lines even if there are 50 chars per line
Three times 33 chars per line would be better for me.

[DMD]

Good morning.
I'm using version 3.5.7, when I load a ".sup" file to create the srt file via OCR, I get this message.

https://i.postimg.cc/SsDVVmYx/Screenshot_001.png



This does not happen with the previous version 3.5.6, how can you solve?
I noticed that in the installation folder, there is a new folder "Tesseract4" instead of "Tesseract", present in the previous version.
Thank you

https://i.postimg.cc/h40hyfGr/Senza_titulo-3.png

[jpsdr]

The last beta linked seems to behave like 3.5.6 with tesseract 3.02.

[DMD]

In version 3.5.7 I have set in the section
Engine mode> Default, based on what is available, it works regularly.
If I select Original Tesseract only (can detect italic) I get the error message.

[mood]

we need a fix version soon to remove Tesseract 4 and folders, to many bugs on Tesseract4

[sneaker_ger]

In the MPC-BE thread (https://forum.doom9.org/showthread.php?p=1854697#post1854697) an mp4 sample was posted. SubtitleEdit (latest beta) handles it differently than ffmpeg and I think ffmpeg is correct.

ffmpeg:
https://i.imgur.com/g8a2zEd.png

SubtitleEdit Beta:
https://i.imgur.com/FbjlpIZ.png

I'm not 100% sure the file isn't broken but maybe you want to take a look at it.

[mkver]

I vaguely remember that there is something odd with ffmpeg and timed text in mp4: The last subtitle would continue until the end of the file. So I would not count (or even bet) on ffmpeg being correct. But I was too lazy to open an issue for it.

[sneaker_ger]

I vaguely remember that there is something odd with ffmpeg and timed text in mp4: The last subtitle would continue until the end of the file. So I would not count (or even bet) on ffmpeg being correct. But I was too lazy to open an issue for it.
https://trac.ffmpeg.org/ticket/6341
?

But this problem is not specifically about the last line.

[mkver]

Yes, that seems to be it. I have already encountered files (which have been created by ffmpeg according to their metadata) whose last line continued for several minutes and for which ffmpeg and SubtitleEdit differed in their timings just like in your screenshots. Therefore they might be related somehow.

[von Suppé]

Hi Nikse,

It seems a few old bugs have sneaked in again in version 3.5.7.

The "Dialogue one hyphen only" plugin also removed the hyphen in a single line subtitle.
Also I encountered several times (not always, though) that, after running "Tools --> Fix common errors", the changes previously made by the "Dialogue one hyphen only" plugin are undone. So I would have to run the plugin again.

In version 3.5.6 I didn't run into these issues.
Can you please take a look?

Thanks

[DMD]

Good morning
I ask if it is possible to enable the font color panel to the hexadecimal code.
The possibility is to copy a color with photoshop in hexadecimal code and the possibility to insert it in the appropriate box.
What I currently do not allow, but only RGB values.
Thank you