This seems like it must be a fairly common captioning requirement:
My sister works for a charity and ran an online seminar (I refuse to call it a webinar!) using Zoom. She had made captions for the 90 minute event by taking machine generated captions and painstakingly correcting them. She planned to put the video & captions up on the charity’s YouTube channel.
The first 40 minutes of the video were not useful and she wanted to trim them. This is easy in YouTube Studio but she found that the captions she had uploaded alongside the video didn’t get trimmed, so all the timings were out of sync.
It seems like a real oversight that YouTube Studio doesn’t do this automatically, but it’s in beta so maybe that feature is coming soon. For now I had to help my sister fix the problem she had spent several hours working on with no progress.
She sent me the VTT caption file which is a text file that looks like this:
00:08:53.340 --> 00:08:54.420
Hi, can you hear me.
00:09:17.550 --> 00:09:18.810
Yes, I can hear you.
I searched DuckDuckGo and found closedcaptionconverter.com, a free online tool for working with captions. It can apply timing offsets so it seemed like the perfect solution… until I tried to apply a negative offset which it didn’t seem to support.
My workaround was to apply a positive offset of 60 minutes minus our required negative offset, and then fix the hours using find & replace.
The first caption after the video was edited needed to started at about
00:00:10.000 but in the caption list it started at
00:35:55.140, so I offset the captions by
Once the captions were offset I fixed the hours by doing a find & replace with regular expression in Sublime Text. I searched for
^01: using the regex symbol
^ to only match characters at the start of a line. I replaced
00:, then replaced
After that I also had to replace
--> 01 with
--> 00, and
--> 01 I sent it back to my sister and… it worked!
One very happy & relieved sister!