OCR Subtitles

[chenlung]

Is Tesseract the best engine for performing optical character recognition on subtitles? Should Tero Subtitler include such functionality?

My experience with soft-coded subtitles:

• SubRip: hasn't been updated for years, only appears to output SRT, and is Windows-only.
• Subler: outputs SRT and is macOS-only.

My experience with hard-coded subtitles:

• Video-subtitle-extractor

[GitGianluc]

@URUWorks how did you get a file like TeroSubtitler/bin/OCR/Spanish.ocr? Did you start up by the relevant Tesseract tessdata file tessdata / spa.traineddata?
I was wondering if and how it is possible to create the .ocr file for the Italian language as well.

[URUWorks]

it was done by hand in early SW versions :)

[GitGianluc]

Wow, I’m not sure if I am capable of producing something similar for the Italian language! I’ll think it over… 🤔

[URUWorks]

everything is possible

[GitGianluc]

Yeah, but I need further information: what is the difference between "Method" : "WORD", "UseRE" : "False" and none of the previous? In other words, what do those options mean?

[URUWorks]

Find: Text to find
ReplaceBy: New text

Method:
WORD: The text is a word
LINE: Find the text on the line
BEGIN: The line begins with the text
END: The line ends with the text

UseRE:
True: Use regular expression
False: Do not use regular expression

[GitGianluc]

Thanks! 😊