Ok, that’s an interesting data point. I know there were some changes to the python venv handling, but this isn’t something I’ve been using myself to know if its newly broken.
It’s possible there’s some conflict with things you’d installed previously, or just something you need to do differently to what you previously did now.
Maybe see Problems configuring Kdenlive for SPEECH TO TEXT and OBJECT DETECTION - #5 by Eugen_Mohr
and if the newly documented procedure doesn’t work for you, file a bug about what you’re seeeing.