Quote:
Originally Posted by Nikse555
SE 3.5.12 is out 
@darksen: Did you get a lot of false corrections? Perhaps that could be improved if you posted some. |
Sorry for taking so damn long to reply back.
Yes, I always get a lot of that. I make subtitles in Spanish and obviously I use the Spanish rules, some errors I get very often are about the OCR correction, sometimes, and this is just one example, SE wants to replace
Marlboro with Mar
iboro (l for i), other times it wants to replace names for their localized ones, like Ivan with Iván, other times it wants to fix quotes when it doesn't have to, in Spanish the main quotes we use are angled quotation marks (« and ») and if there is a quote inside that quote we use double quotes (" and ") and if there is another quote inside that we use single quotes (' and ' ) but SE sometimes wants to fix the double quotes and replace them with the angled ones despite they being inside angled already, i.e:
Code:
Ivan said: «Blahblahblah "blah-blah-blah" blahblah»
SE wants to fix it like:
Code:
Ivan said: «Blahblahblah blah-blah-blah blahblah»
Or sometimes:
Code:
«Blahblahblah «blah-blah-blah» blahblah»
It also happens when the dialogue is split in more than one subtitle because SE doesn't detect that angled quotes have been used in the previous sub.
And there have been more errors but I can't remember them all right now. That's why I'm asking to split the "Fix common OCR errors (using OCR replace list)" into two, one informing that the default list is used for said correction and the other that the correction comes from the user list.
Thank you.