as stated in the title, I just created my first Croatian lesson from an audio file, using the automatic transcript feature. The text created uses the Cyrillic script. It should be in the Latin script, though. How can I fix this?
LingQ doesn’t expose any configuration options for the Whisper API.
To work around it you would need to transcribe the file somewhere else. You will also need a model that allows prompting if the responses flip flop between Latin and Cyrillic for Croation. GPT is probably the way to go, but can be pricey.
That’s unfortunate. I installed openai-whisper on my machine and by specifiying the language I got a perfectly fine translation written in the Latin script in less than 10 minutes. But Lingq can’t pass the language option? What a shame.
Is there any effort put into fixing this issue? I just got part of an audio file transcribed with the Latin alphabet, and part of the same file with the Cyrillic alphabet.
The language of the course I’m talking about is set to Croatian, not to Serbian. Why are we talking about Serbian? The two languages might be very similar, but Croatian is not written with the Cyrillic alphabet!