Help defining workflow for sub retiming

[Filippo]

Hello,
I have two .srt files in two different languages, EN and IT. One has the correct timing, the other needs to be adjusted. The two subs have different number of lines, the IT has less but because they are merged into one.

I'm using Subtitle Editor, Is there a way to port the EN timing to the IT sub without doing it one line at time? I tried with visual sync but the result is bad because the movie goes out of sync fast. So defining 3 or more points is not useful because all the lines in between are unaligned. The movie is in japanese so after a while I can't really tell if the line is matching the actual speak, I can't really rely on audio feedback.

Is there a tool (maybe ai based?) that could match EN-IT lines using the EN times?

Thank you

[CDK]

You can try this https://blu-disc.net/download/STA.7z (Windows only)

[pcspeak]

I ran an online experiment using ChatGPT.
Firstly I needed to break up the srt file and create 7 separate srt files containing no more than 80 timing numbers as I hit token limits. The show has 555 timing numbers.
e.g.
1
00:00:24,080 --> 00:00:26,640
Good evening, and welcome
to a very special edition

2
00:00:26,640 --> 00:00:28,280
of previously unseen clips

3
00:00:28,280 --> 00:00:30,800
from series 18 of "Would I Lie To You?".

4
00:00:30,800 --> 00:00:34,080
Joining David Mitchell tonight - Stacey Solomon,
.
.
.
.
.

80
00:04:14,840 --> 00:04:19,160
OK, so I went to the Met Gala a few years ago...
Ooh!

I queried ChatGPT:
Create a new video subtitles file, converting the following subtitles from English to Italian. Keep the timing.

After 'Keep the timing' I copied and pasted the content of the first of the 7 cut srt files. On the same line. I seems a bit odd, but the AI software had no problems.


This is the result:

Here is the translated subtitle file in Italian, preserving the original numbering and timing:
1
00:00:24,080 --> 00:00:26,640
Buonasera, e benvenuti
a un'edizione davvero speciale

2
00:00:26,640 --> 00:00:28,280
di clip mai viste prima

3
00:00:28,280 --> 00:00:30,800
dalla serie 18 di "Would I Lie To You?".

4
00:00:30,800 --> 00:00:34,080
Con David Mitchell questa sera - Stacey Solomon,
.
.
.
.
80
00:04:14,840 --> 00:04:19,160
OK, sono andata al Met Gala qualche anno fa...
Oooh!

After querying ChatGPT 7 times I would have the Italian translation. I would need to combine all 7 files into 1, and check the timing number sequence is correct.
All in all I thought the result was very good.
It's a little fiddly breaking the srt into 7 separate files. I should have only done the first 2 as I only posted 2 to ChatGPT.

I hope this helps.
Cheers.

[pcspeak]

Using Subtitle Edit - Tools/Split Subtitle - Number of equal parts
I set the parts count to 8 and, Voilà!