I wanted to rip hardcoded subtitles from movies and turn them into to an SRT file so that I can select the subtitle and look up words I don't know.
I saw there were four parts to this problem.
-
I needed a program that could get me screen grabs of each subtitle as well as the time info for said subtitle.
I used VideoSubFinder https://sourceforge.net/projects/videosubfinder/
-
I needed a program to clean up the screen grabs.
I wrote one
-
I needed a program to OCR the screen grabs.
I used tesseract https://github.com/tesseract-ocr/tesseract
-
I needed a program to build the SRT from the results of OCR.
I wrote one
What is this even? This is my solution to this problem.
Video Sub Finder is a program written by Simeon Kosnitsky. It takes screengrabs of subtitles and saves them with a name consisting of the time the subtitle first appears and the time it disappears. This is the solution i used for steps 1 and 2.
Tesseract is an Open Source OCR engine owned by google. This is part of the solution i used for step 3.
resize_images.py was written by me This is the solution I used for step 2.
build_subs.py was written by me. This is the solution I used for steps 3 and 4.
How to use these scripts?
Video Sub Finder will create a folder called TXTimages, put both scripts in this folder.
After you have generated the TXTimages then run resize_images.py, after that's done run build_subs.py Running them in either idle or CMD is best as you can see their output.
Change the variables at the top of each script to suit your liking, I have left them on the settings i found most useful. In resize_images.py
I apologise in advance to the poor soul who tries to read or use this code, I wrote it when I was tired and a little bit drunk. These scripts have the added feature of turning your computer into a heater for the duration of their use.
Install dependencies
run pip install numpy PIL opencv-python pytesseract
Dependencies:
- python 3
- opencv
- pytesseract
- PIL
- Numpy