pajowu

Thank you for reporting this issue. To properly investigate it, could you please send us the debug log. You can export it from the "Help" menu -> "Export debug log"....

Should be fairly easy by removing the model caching in https://github.com/audapolis/audapolis/blob/ff91a2c23c31a8179c3d12f348eb91431d0dfb2b/server/app/models.py#L119-L126 Feel free to open a PR, else we'll so it once we get to it

Some more explanation on the current behaviour: We thought about how we want to handle model loading. Since the model loading take a while, we said we didn't want to...

I’ll test it on my devices before the merge. Can you fix the CI?

It is possible to edit every word by placing the cursor after it and pressing i (or placing the cursor in front of the word and pressing o). If you...

Yes, there is. We need to keep the mapping word-text timing. Allowing one to edit the entire document as if it were a normal text would make that difficult

Also I don't think always editing the transcript is a good decision. While editing the transcript text is an important part, I think the main focus of this app is...

> Without an accurate transcript, I'm not really sure how useful this app is. We agree, that's why we currently have a word-/selection-based editing, which allows users to quickly edit...

> Okay then, some ux ideas: We had these idea as well. Some are already on our wishlist, but couldn't be implemented yet because of missing time, for some there...

Two more things: 1. We recently added a special mode which highlights low-confidence words. As with most transcript correction features I talk about in this thread, you need [Version 0.2.2](https://github.com/audapolis/audapolis/releases/tag/v0.2.1%2B39)...