Timecodes past 1 hour mark written incorrectly
I noticed while checking the titles uploaded for the Walz Colloquium video that subtitles stopped near the end of the video. As it happened, this occurred almost immediately after the hour mark.
Captions past the 1 hour mark are written as such:
1081
00:60:55.579 --> 00:60:58.440 A:middle L:90%
you for being here. Thank you. Okay.
In this particular example (from another video) the minutes fields in the timestamps have the value "60". However, values greater than 59 are invalid in this field; the hours field is meant to be used for videos longer than 60 minutes, with the minutes field rolling back over to zero, as you might expect.
I've fixed the invalid timestamps in the Walz Colloquium video.
The code from the project that I found to convert the AWS json files to vtt files did not take into account videos longer than one hour. I've made a change to write correct timestamps after the one hour mark, and am currently testing this change.
After this change, I expect that Caption Whale will work correctly with videos up to 4 hours long and 2Gb in size (I believe these to be the AWS transcribe limits).