cxdy / srtlate Goto Github PK

I forgot how languages work when building this.

Is there TL;DR for too long, didnt type? If not, this will be it's birth.

Anyways, TL;DT
As of the current version, each individual subtitle is translated out of context, so the translations don't necessarily make sense. I only (and barely) speak English, so I have no idea exactly how bad they are, but we need to figure out a way to determine if subtitles are related to each other, and then translate them so that they make some kind of sense.

IDEAS:

• Build the SRT file through the program and run it through multiple languages at once
  This is probably the most difficult approach, but also probably our best bet (and it's cool as fuck). Only thing is, I don't know how to do that yet. If we can figure it out, maybe there's a way to tell if a different person is speaking?

• Determine if the times in the previous & next subtitle are close
  00:00:11,636 --> 00:00:13,221
  I'm assuming the format is HR:MIN:SEC,MS? I'm just not 100% what the number after the comma is - seems like a safe assumption though. I know we talked about it before but I forgot.
  Anyway, I guess for this approach, we could just create an empty list and add the subtitles that start within a second or two of the previous subtitle, run it and see if it makes sense? Only thing is, I don't know how we'd determine if it makes any sense with a computer. I can't speak other languages so I'm not confident that I can teach a computer how to speak other languages.

As of the current version (April 28 2019), we start with an English (or whatever language, I guess) SRT file. The file is converted to JSON so we can interact with the file programatically.

Line breaks from the source SRT file are converted to \n in the JSON file cause JSON, and when the program goes to translate the captions, we pull the line break because obviously \n won't translate. Gotta figure out a way to add them back though.

@kpmgeek mentioned that in America, the standard for each line in an SRT file is 32 characters, but other countries are more lenient with their standards. I think we can just count up to 32 characters, and if there's a character that is not a space (i.e: a letter, number, etc), we move back to the nearest space and insert a line break there.

However, this isn't necessarily the best approach because sometimes on the first line you have things like [Narrator] and then the second line is what the narrator is saying, so we need to find a way around that & also find a way around leaving one word/character/whatever on the 2nd line with everything else up top. Maybe count the total characters for the caption (between 1-2 lines), then count all the spaces in that caption, subtract the number of spaces from the number of total characters and divide that number by 2?

Example Caption:
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Etiam et maximus dolor.
Total Length - 80 characters
Number of spaces - 9 spaces
(80 - 9) / 2 = 35.5

I don't know where I'm going with that, but you get what I mean.