Settings for ignoring sequences of languages not known to whisper

[mojo12344]

Hi, I have a lot of transcripts containing a mix of interview questions asked in English, replies in languages not known to whisper and translations done by research assistants. noScribe seems to get very confused by the unknown languages - is there a way to tell the model to ignore those sequences it does not recognize?
Cheers!

[kaixxx]

That's a difficult case. You can try the "multilingual" setting in the language field. This will still try to transcribe the unknown language, but might help to get cleaner English parts. Otherwise, I have no idea...

[mojo12344]

Thanks a lot, I'll give it a try!