It seems that the auto generated english content for audios & videos is not very accurate, I found a lot of mistakes and missing content in it, can you please help to do some improvement on it?

BTW. I also use whisper with '--model medium' option on my own laptop to generate the text for audios. actually, it a little more accurate than the lingq platform.

If you think we can not improve it by upgrade model level, if it's possible to allow the users upload the text for audios by themself? I think maybe they already have those content which is more accurate.

Thanks a lot! I love LingQ, undoubtedly, it's a great platform!