Michael Wuttke

@lkiesow: Will you create a PR for it?

Oh. Sorry, I thought the PR only concerned logging.

With whisper-ctranslate2 I haven't had any success with our setup (Opencast 14.9 & whisper-ctranslate2 [0.3.9](https://github.com/Softcatala/whisper-ctranslate2/releases/tag/0.3.9)) so far. openai-whisper ([v20231117](https://github.com/openai/whisper/releases/tag/v20231117)) with OC 14.9 works on the other hand as expected.

I have successfully tested the current status of this patch with the Opencast Version 15.2 together with the language packs that @gregorydlogan has built. Would it be possible to merge...

I have only changed the enginetype in `/etc/org.opencastproject.speechtotext.impl.SpeechToTextServiceImpl.cfg` to whispercpp. That was all. After that, the transcription ran without errors.

We have also been able to detect the error in our OC system with large videos (Videos 90min / 2 GB (Cam) + 200 MB (Slides) / MP4). **Server environment:**...

> What happens between those two lines? > > ``` > 2024-04-11T09:59:32,096 | DEBUG | (WhisperCppEngine:362) - > ... > 2024-04-11T11:19:22,512 | DEBUG | (WhisperCppEngine:362) - output_vtt: failed to open...

> Can you run just this whisper.cpp command on this box? > > ```shell > whisper.cpp --model /usr/share/whisper.cpp/medium.bin/ggml-medium.bin -ovtt -oj --output-file /srv/opencast/workspace/collection/subtitles/job-12146265/HU_LCP_007_20221021_091000_S2R1 -t 8 --language de /srv/opencast/workspace/mediapackage/a225fa83-3181-4b78-b169-e2a5a904607b/d323609e-6f30-4e4b-aa4a-e1f2aeebf64c/HU_LCP_007_20221021_091000_S2R1.wav > ``` >...

Hello @gregorydlogan. The transcription runs cleanly to the end. As you can see from the logs. I just hadn't copied it. The mount point is also retained until the end....

A successful example for a possible implementation for virtual camera backgrounds or the blur effect is BigBlueButton, I guess.