Mirrors/SubtitleEdit
1
0
Fork 0
mirror of https://github.com/SubtitleEdit/subtitleedit.git synced 2024-11-22 03:02:35 +01:00
Code Issues Projects Releases Wiki Activity

Improve ocr dictionaries slightly

Browse Source
This commit is contained in:
Nikolaj Olsson 2020-05-07 07:57:02 +02:00
parent 1ef81526cf
commit 77f98581ff
4 changed files with 3397 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

112
Dictionaries/eng_OCRFixReplaceList.xml
View File

@ -2656,6 +2656,112 @@
<Word from="ZSth" to="25th" />
<Word from="d0n't" to="don't" />
<Word from="D0n't" to="Don't" />
<Word from="anyjobs" to="any jobs" />
<Word from="Sogreat" to="So great" />
<Word from="hearfromyou" to="hear from you" />
<Word from="meansyou're" to="means you're" />
<Word from="forjust" to="for just" />
<Word from="notjoking" to="not joking" />
<Word from="hidbehind" to="hid behind" />
<Word from="Itstarts" to="It starts" />
<Word from="outlike" to="out like" />
<Word from="reallyfast" to="really fast" />
<Word from="Andthe" to="And the" />
<Word from="andit's" to="and it's" />
<Word from="completelyalone" to="completely alone" />
<Word from="wonderedifanyone" to="wondered if anyone" />
<Word from="changedevery" to="changed every" />
<Word from="reallyjuvenile" to="really juvenile" />
<Word from="lovedher" to="loved her" />
<Word from="thingsyou" to="things you" />
<Word from="Butmyheartache's" to="But my heartache's" />
<Word from="Didyou" to="Did you" />
<Word from="standbyme" to="stand by me" />
<Word from="notatall" to="not at all" />
<Word from="Nowdeparting" to="Now departing" />
<Word from="haveyourattention" to="have your attention" />
<Word from="IosAngeles" to="Los Angeles" />
<Word from="haveyourtickets" to="have your tickets" />
<Word from="Atlast" to="At last" />
<Word from="Mylove" to="My love" />
<Word from="walkedto" to="walked to" />
<Word from="herapartment" to="her apartment" />
<Word from="intoxicatedby" to="intoxicated by" />
<Word from="thepromise" to="the promise" />
<Word from="wouldalign" to="would align" />
<Word from="Whatjust" to="What just" />
<Word from="oftheyearare" to="of the year are" />
<Word from="Theybegin" to="They begin" />
<Word from="andtheyend" to="and they end" />
<Word from="IfTom" to="If Tom" />
<Word from="hadlearnedanything" to="had learned anything" />
<Word from="thatyou" to="that you" />
<Word from="cosmicsigniflcance" to="cosmic signiflcance" />
<Word from="allanything" to="all anything" />
<Word from="everis" to="ever is" />
<Word from="wasprettysure" to="was pretty sure" />
<Word from="grewup" to="grew up" />
<Word from="untilthe" to="until the" />
<Word from="totalmisreading" to="total misreading" />
<Word from="Thegirl" to="The girl" />
<Word from="didnotshare" to="did not share" />
<Word from="herlong" to="her long" />
<Word from="darkhair" to="dark hair" />
<Word from="Andfeelnothing" to="And feel nothing" />
<Word from="slightlyabove" to="slightly above" />
<Word from="Forallintents" to="For all intents" />
<Word from="andpurposes" to="and purposes" />
<Word from="justanothergirl" to="just another girl" />
<Word from="schoolyearbook" to="school yearbook" />
<Word from="Colormylife" to="Color my life" />
<Word from="Thisspike" to="This spike" />
<Word from="oftheiralbum" to="of their album" />
<Word from="topuzzle" to="to puzzle" />
<Word from="theirlives" to="their lives" />
<Word from="finditnow" to="find it now" />
<Word from="commercialbuildings" to="commercial buildings" />
<Word from="andyours" to="and yours" />
<Word from="anypity" to="any pity" />
<Word from="Areyou" to="Are you" />
<Word from="armyis" to="army is" />
<Word from="caughtsome" to="caught some" />
<Word from="cityofthe" to="city of the" />
<Word from="Didhegive" to="Did he give" />
<Word from="everyonego" to="everyone go" />
<Word from="evilandgood" to="evil and good" />
<Word from="findthe" to="find the" />
<Word from="flndthem" to="flnd them" />
<Word from="Forbeing" to="For being" />
<Word from="greatnumbers" to="great numbers" />
<Word from="heardthe" to="heard the" />
<Word from="helltheyare" to="hell they are" />
<Word from="hisprivate" to="his private" />
<Word from="Howdo" to="How do" />
<Word from="inflictmore" to="inflict more" />
<Word from="inyourinflnite" to="in your inflnite" />
<Word from="mydear" to="my dear" />
<Word from="ofwar" to="of war" />
<Word from="orgreed" to="or greed" />
<Word from="orlust" to="or lust" />
<Word from="ourhands" to="our hands" />
<Word from="ratherprimitive" to="rather primitive" />
<Word from="realnice" to="real nice" />
<Word from="talklike" to="talk like" />
<Word from="That'llbe" to="That'll be" />
<Word from="thatpowerhere" to="that power here" />
<Word from="Theyare" to="They are" />
<Word from="topossess" to="to possess" />
<Word from="toyou" to="to you" />
<Word from="wantpeace" to="want peace" />
<Word from="we'dbe" to="we'd be" />
<Word from="whichyou" to="which you" />
<Word from="Whycan't" to="Why can't" />
<Word from="willlook" to="will look" />
<Word from="willmake" to="will make" />
<Word from="willmurderhis" to="will murder his" />
<Word from="wishyou" to="wish you" />
<Word from="wouldnot" to="would not" />
<Word from="yourpeople" to="your people" />
</WholeWords>
<PartialWordsAlways>
<!-- Will be replaced always -->
@ -2725,6 +2831,7 @@
</WholeLines>
<PartialLinesAlways>
<LinePart from="forbest act" to="for best act" />
<LinePart from=",.," to="..." />
</PartialLinesAlways>
<PartialLines>
<LinePart from=" /be " to=" I be " />
@ -3045,6 +3152,10 @@
<LinePart from="You' re" to="You're" />
<LinePart from="you' re" to="you're" />
<LinePart from="You' ve " to="You've " />
<LinePart from="he 'd " to="he'd " />
<LinePart from="ofb" to="of b" />
<LinePart from="ofM" to="of M" />
<LinePart from="ofS" to="of S" />
</PartialLines>
<BeginLines>
<Beginning from="lgot it" to="I got it" />
@ -3197,6 +3308,7 @@
<Ending from="pshycol" to="psycho!" />
<Ending from=" i..." to=" I..." />
<Ending from=" L." to=" I." />
<Ending from=" ." to="." />
</EndLines>
<RegularExpressions>
<RegEx find="([a-z]) Won't " replaceWith="$1 won't " />

3282
Dictionaries/eng_OCRFixReplaceList.xml.bak Normal file
View File

File diff suppressed because it is too large Load Diff

BIN
Ocr/Latin.db
View File

Binary file not shown.

6
src/Forms/Ocr/VobSubOcr.cs
View File

@ -5583,13 +5583,13 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
return false;
}
line = line.Replace("[", string.Empty);
line = line.Replace("]", string.Empty);
line = line.RemoveChar('[');
line = line.RemoveChar(']');
line = HtmlUtil.RemoveOpenCloseTags(line, HtmlUtil.TagItalic);
var arr = line.Replace("a.m", string.Empty).Replace("p.m", string.Empty).Replace("o.r", string.Empty)
.Replace("e.g", string.Empty).Replace("Ph.D", string.Empty).Replace("d.t.s", string.Empty)
.Split(new[] { ' ', '.', '?', '!', '(', ')', '\r', '\n', '\t' }, StringSplitOptions.RemoveEmptyEntries);
.Split(new[] { ' ', ',', '.', '?', '!', '(', ')', '\r', '\n', '\t' }, StringSplitOptions.RemoveEmptyEntries);
foreach (string s in arr)
{
if (s.Length == 1 && !@"♪♫-:'”1234567890&aAI""".Contains(s))