Subtitleedit: OCR issue with command line

[svb91]

Hello!

It is not possible to specify the engine mode of the OCR feature while using SubtitleEdit via comand line.
SubtitleEdit use "Tesserac only" and not my default value "Tesserac + LSTM".
Therefore many subtitles are useless because it does not recognize the text correct.

Is there any possibility to fix this issue or is there any additional parameter which I can set in the cmd to use LSTM for OCR?

[xylographe]

It looks like Subtitle Edit uses Configuration.Settings.VobSubOcr (see src/Logic/CommandLineConvert.cs).
Those are the OCR settings that were used previously, as stored in in the <VobSubOcr> section
of your Settings.xml file (in your %AppData%\Subtitle Edit folder).

[andiandi13]

What does that mean ?

I'm facing the same problem.

I want to convert 100 sup files to srt, with both tesseract method and LSTM method in two separate batch.

How can I achieve it ?

[xylographe]

@andiandi13
I suppose, the most convenient way would be to,

• exit Subtitle Edit
• open the %AppData%\Subtitle Edit folder in Explorer
• copy Settings.xml to Settings.GUI.xml
• start Subtitle Edit
• select "File > Import/OCR Blu-ray" to open a sup file
• set up the OCR parameters (tesseract etc.)
• click OK to close the Import/OCR dialogue
• exit Subtitle Edit (decline when asked to save the empty subtitles)
• rename the changed Settings.xml to Settings.OCR-tess-LSTM.xml
• rename Settings.GUI.xml to Settings.xml to restore the original settings

Alternatively, you could edit (a copy of) the Settings.xml file manually in a text editor, like vim or notepad. In this case you would find the OCR parameters in the <VobSubOcr> section.

Run the command line conversion from within a script, that copies Settings.OCR-tess-LSTM.xml to Settings.xml before invoking SubtitleEdit.exe. In a powershell script, for example,

Copy-Item -Destination "${env:APPDATA}\Subtitle Edit\Settings.backup.xml" -LiteralPath "${env:APPDATA}\Subtitle Edit\Settings.xml"
Copy-Item -Destination "${env:APPDATA}\Subtitle Edit\Settings.xml" -LiteralPath "${env:APPDATA}\Subtitle Edit\Settings.OCR-tess-LSTM.xml"
SubtitleEdit.exe /convert @args | Write-Output
Copy-Item -Destination "${env:APPDATA}\Subtitle Edit\Settings.xml" -LiteralPath "${env:APPDATA}\Subtitle Edit\Settings.backup.xml"

[darnn]

Or obtain the desired settings.xml file as described here, and just extract the portable version of SE to a different folder and use that settings file with it.

[andiandi13]

I have portable Subtitle edit so I can replace Settings.xml with preferred OCR method.

But now, how do I do a batch command line (no GUI) ?

When I run a batch conversion in the GUI, it seems that it'll never finish one file, I get impatient and I close it (but when I OCR only one .sup file, there's no problem)...

[andiandi13]

Never mind, I updated Subtitle edit to the last version and the batch now works.

But I still wonder, can we do a command line batch in Windows (.bat file) ?

[xylographe]

Run SubtitleEdit.exe /? or SubtitleEdit.exe /help to show the supported command line parameters.