Subtitle Edit OCR Doesn't Recognise Capitals for W, O, V, U, etc.

[koberulz]

So I end up with subtitles like:
"MoTHER: <i>welcome.</i>"
"<i>(THE ENTRY OF THE GODs
lNTO vALHALLA</i> PLAYING)"
"DAVID: (AS wALTER)"

Etc.

Is there anything I can do to fix it?

[johns0]

Uncheck auto-correct and fix them,it's a pain but that's the only way to get reliable results.

[koberulz]

Well, yeah, that's the obvious answer but I don't really fancy manually editing tens of thousands of subtitles to be honest. There's nothing that can make the OCR work better? I'd expect the occasional mistake, sure, but it just refuses to count any of those letters as capitals, ever.

[johns0]

If your editing thousands of subtitles hire other people cause that's too many subs for anyone to work on.

[koberulz]

Is there a movie that doesn't have thousands of subtitles?

[johns0]

A movie has one subtitle with thousands of words,not the other way around.

[lingyi]

You may be able to speed things up by setting everything to lowercase, then sentence case in Word or equivalent.

[koberulz]

Tesseract 3.02 gives far better results than the default binary image compare, although it tends not to catch spaces at times. Significantly fewer overall errors though.