[REL] Negerkys og labre larver (1987)

[ghost]

So why don't you use the dictionary? It saves you a lot of time afterwards. I never let it just run without interrupting. There are too many errors in the result.

About speed: I guess it depends of the power of your CPU how fast it*s running.

And yes, you're right: Why did I extract the idx/sub?

[Night457]

So why don't you use the dictionary? It saves you a lot of time afterwards.

Mmmm ... I'm not sure. I did have it set to use the downloadable Danish dictionary, so that it would recognize most Danish words and not assume they were English. If you mean the box on the right that suggests possible choices for questionable words, then I HAVE also used that, many times. But there are occasions where the OCR totally fails to recognize a character (italics are a problem) and I just have to type out a character, or a word, or sometimes even edit the entire line.

I never let it just run without interrupting. There are too many errors in the result.

Absolutely! That is what I meant by "too many" errors. I will only actually use OCR results if they have been monitored. I only did it on autopilot as a test run because I was not fool enough to volunteer for a job that might be unexpectedly difficult! (Like OCRing hardcoded subtitles. Eww.) I looked at the test run and decided that it was at least good enough that I could handle doing it properly.

Remember, I am one of those nutters who can watch and enjoy a movie that does not have any subtitles, even though I prefer having them. popdrome said he understood most of it, and he has not indicated that Danish is one of the languages he knows well. (Though maybe it is closer to Dutch than even *I* thought.)

About speed: I guess it depends of the power of your CPU how fast it*s running.

I read about your new machine, so I know yours is far faster!

Also, speed is largely dependent on my increasingly poor vision and my tiny monitor. I have to look closely at the questioned text to make sure I get it right, especially with any language that is not English.

And yes, you're right: Why did I extract the idx/sub?

Because ... reasons?

Maybe because idx/subs are much smaller than MKVs so it will process faster during the OCR? I would have to compare to see. And once I have OCRd a subtitle once, I don't want to do it all over again just for a speed test!

I decided to challenge myself to see how my OCR speed would compare to ghost's estimate of his taking about 10 minutes. With my poor vision, tiny monitor and slow computer it took me about 25 minutes. This is still much faster than I expected originally. I used Tesseract 5.00 on an MKV. The vast majority of the "errors" were when it would not recognize English words using the Danish dictionary. Maybe ghost is clever enough to get it to use multiple dictionaries, or in previous Danish OCR he has already added those English words to the Danish dictionary so that it recognized them this time around.

Well, there I went thread crapping boring tech into a movie Release thread ... again. I'm about to go watch the movie now!

[ghost]

I looked at the test run and decided that it was at least good enough that I could handle doing it properly.

OK, and what do you do with your finished file? How do you know, how it is written correctly? During the OCR you have control and can compare each letter with the picture of the subtitle and can easily correct it. Just my opinion, I prefer it that way.

Generally the dictionary works pretty well when I use it.

Maybe because idx/subs are much smaller than MKVs so it will process faster during the OCR?

No, I think that doesn't make any difference. SE extracts the subs temporary and is loadig them (at least I think so...).

'Nough said. Let's enjoy the movie.

[Night457]

OK, and what do you do with your finished file?

I looked at it, and then I saved it in case I wanted to look at it again.

How do you know, how it is written correctly?

I don't need to know Danish to see some cases where it is NOT correct, from looking at the results, like

V/

1V 2V 3V 4V 5V
6V 7V 8V 9V 10V

That is just nonsense.

(which should be
Vi

1, 2, 3, 4, 5,
6, 7, 8, 9, 10)

Or if a translation is gibberish, then the original probably is also. Like if you take a Youtube automated speech-to-text Russian file and translate it to English, some of it will be really weird, because it gets the Russian original wrong.

During the OCR you have control and can compare each letter with the picture of the subtitle and can easily correct it. Just my opinion, I prefer it that way.

That is how I do it! I agree with you! I only did the automated run because it takes no effort and gives me a quick idea of how bad doing it properly would be! I machine-translated the automated Danish OCR to English, and it was still mostly comprehensible, which I could see right away scrolling through with a few minutes reading. That means that while there would be errors in the OCR, there would not be A WHOLE LOT of them. So fixing them would not be hard. It only took about 10 minutes of my actual time on the computer converting ISO to MKV, loading MKV into Subtitle Edit, running an automated OCR in SE, looking quickly at the Danish, machine-translating it in SE to English, and then skimming through that English file to see how good or bad it was. During the computer processing time I went and did other stuff, and came back later for the next step. I watched stuff on the TV and ate brunch during that time. I could have kept my laptop with me if I wanted and did it that way, too. As I said, it usually takes me an hour or more for OCR, so a 10 minute trial run is nothing compared to that. You, on the other hand, already have the experience and confidence to know that you can handle whatever any subtitle throws at you. So you "Just Do It" as the sneaker ads said.

'Nough said. Let's enjoy the movie.

Yeah, I did that!

THANK YOU POPDROME AND THANK YOU GHOST!!!

WIth the subtitles I completely understood the dialog, which was pretty funny at times. Still, the plot left me bewildered: I had no idea what was going on much of the time. This was a totally Eighties bizarro world. I enjoyed it!

[popdrome]

THANK YOU TOO

Speaking of gibberish, a swede would say that Danish ofcourse *is* gibberish. The way they talk man, like babies!
No disrespect

I'd say if you listen closely it resembles english more than dutch, whatever these are al Germanic languages so..
I totally loved this film. It was way too short and the Prince impersonator (who is actually Michael Jackson) is HILARIOUS. OMG the girls are so cute...

Thanks for the subs ghost! Here's my payment:
[Image]

[Ziggy Plock]

I may have a hard time distinguishing Danish from Dutch

There's a difference????

So far I have just done quick automated OCR without manual corrections (...) I would be happy to do monitored OCR with live corrections so you can do something more challenging, like OCRing hardcoded subtitles on another movie

I've a got a Chinese movie with hard-coded KOREAN subtitles ... in white cursive letters (Luckily it's sans serif, so more like Arial/Verdana and not like Times Roman/Courier) ... if anyone's interested??
https://www.imdb.com/title/tt7232860/

Like this post to see ed2k links  [794.45 MiB]

[jonas]

And yes, you're right: Why did I extract the idx/sub?

https://subtitletools.com/convert-sub-idx-to-srt-online

[ghost]

And yes, you're right: Why did I extract the idx/sub?

https://subtitletools.com/convert-sub-idx-to-srt-online

Interesting. I'll try it next time.

Thanks for the subs ghost! Here's my payment:

I got flowers

[jonas]

And yes, you're right: Why did I extract the idx/sub?

https://subtitletools.com/convert-sub-idx-to-srt-online

Interesting. I'll try it next time.

Sometimes you have to wait in line. But it works 100 percent. Tested.

[pillowbaker]

I'd like to add my voice to all the thanks going on here. Much appreciated for the subs, ghost, and thanks to everyone involved in extracting them and getting the movie shared. I'm excited for this crazy 80s romp (with an MJ cameo, no way!).