Random strings and wrong special characters in edited subtitle text

[Piriya-VM]

We use Vosk to create subtitles and edit them in the subtitle editor.
After saving and publishing, the subtitle file includes strange strings and special character like ä, ü, ö don't get displayed correctly.
See the attached vtt file, screenshot and partial-publish Workflow.

WEBVTT

3lPSCXPUvSk3QfzzsrXDj
00:00.420 --> 00:05.930
Apfel

iOmyuq7YoRVkBdHrDHj8m
00:06.930 --> 00:10.530
Hallo ich bin Vosk und versuche nur 
zu Ueberleben irgendwas irgendwas.

3bW92IwxDIrqu0k_HdIeu
00:10.530 --> 00:14.100
orange

zPaQJaT9U0iBY2JI5sqFS
00:14.100 --> 00:18.750
kiwi 
irgendwas irgendwas irgendwas

lBh9Dx0GfstMMqgq-R1XG
00:19.920 --> 00:22.860
ananas

partial-publish.txt

[DZenker]

I can confirm this bug, this also happens in our installation (OC 12.6).

[fsufrre]

I can confirm this bug too. (OC 12.4)
Any ideas when this bug is fixed?

[Arnei]

Interestingly, I cannot confirm this bug (OC 0fb1c14, also newer versions). The characters ä, ü, ö render normally.

The "strange strings" like 3lPSCXPUvSk3QfzzsrXDj are subtitle cue ids generated by vosk. I do not know why vosk generates them, but looking at the downloaded subtitle file they are perfectly legal WebVTT. Maybe Paella Player doesn't know how to handle them?

[fsufrre]

we are using amberscript. but i think the way of delivery is not the point. while saving the edited subtitle and republish the transcription, the *vtt file gets some extra strings. its visible in paella and also after downloading the *vtt File. I think the magic happens in "saving" process.

[Arnei]

True, it seems blaming the strage strings on vosk was hasty on my part. They are still legitimate WebVTT cue ids though, so maybe that's worth a separate issue in the main Opencast repo on how Paella handels cue ids.

[lkiesow]

This should be fixed in recent versions of Opencast by now.
For more details, see opencast/opencast#4764

If the problem persists regardless, please re-open.