Just an update to a problem discussed in the regular forum. See this thread for reference:
This appears fixed in 22.1.1 Build 7. I did not test heavily, but two variations that did not work before get correct results now: pick in to out for the whole sequence or set an out point only.
This problem is back in the 22.4.0 Release version, and today I confirmed it is present in Beta 126.96.36.199. I have at least 6 different threads where the problem is reported. It is an easy workaround, but still....
Post in regular forum:
"Your transcript shows that it starts at the 8 HOUR mark. This was a bug that started about 22.0.0 and was fixed in 22.1.1. Apparently some version of it is back.
As reported by @5C97, the workaround is to turn OFF transcribe in to out, or as also discovered in the past version of this bug, make sure that both an in AND out marker are set when using transcribe in to out. When you don't set an actual in or out marker, even though it acts like the beginning is an in point and the end is an out point, they are not. (There is no in or out point to "clear," for example.)
See this previous thread:
@NicoBecherer Has anyone reported this for a new bug report?"
I'm still having this issue with 22.4. I've tried everything in the thread and nothing seems to be fixing the incorrect timecode timing for my speech-to-text!
ruth, please provide a screenshot of the transcription tab with the transcription complete. The problem discussed here is avoided by a) turning OFF transcribe in to out or b) adding BOTH an actual in point and an out point (not using the beginning as an in point or the end as an out point).
If that does not work, you have some other problem.
Note that this is the Beta forum. I last tested and found this problem in Beta 188.8.131.52. I see that today, the Beta build is 184.108.40.206. I have not tested there yet.
In PR Beta 22.6.0 Build 5, there is a change, that appears to prevent the problem by a) disabling the checkbox for "in to out" if there is neither an in or out point set and b) proceeding as if there is an automatic in point set if there is only an out point set,and vice versa. I can't tell if there is actually an automatic in or out point set; one is shown in the source monitor, but that always happened before anyway.
In any event, I would consider this a fix, since a) you shouldn't be able to specify in to out if you are transcribing the whole sequence and b) it now works if only in or out is set by the user.