Error encountered while transcribing low resource indian languages - Malayalam and Assamese

[shubzk]

While transcribing Malayalam and Assamese audio by using the large-v3 model, the transcription generated is in Punjabi. Help me with a solution if you have encountered the same with any other languages.

[jongwook]

I'm afraid Whisper is not practically usable for Malayalam or Assamese. Similar discussions regarding Amharic in:

• Language detected in 30 seconds, transcript returned in another language's character set. #1735

[nairajay2k]

Any roadmap to fix for malayalam etc?