Wrong language detected

#24
by delectacion - opened

When trying to transcribe this video (https://www.youtube.com/watch?v=NNZkPi2Ruv8) the transcription comes up in Catalan instead of Spanish. Is there a way to change the transcription language in the demo? If not, would you consider integrating it? I've never had this problem before and I'm not sure what is happening with this video.

Thanks for you work!

@sanchit-gandhi Sorry to insist, but I've encountered the same problem with more videos. I just transcribed a video (full audio in Spanish) that gets parts transcribed in Catalan or Galician for some reason. Do you think a language selector could be added to the Space? I tried using the kaggle version but I'm too code-ignorant to get it to work. This is certainly the fastest Whisper implementation I've come across, so I'd love to see that functionality in this demo.
Here's an example of the transcription I got (red Catalan, blue Spanish, green Galician).
mixed languages.png

Sign up or log in to comment