The Text-To-Speech extension does sort of work, but not usually. If I try to have it say one word (like “Hello”) it's usually fine, but when I type more than one word (like “Hello to the world”), it will rarely say anything. And sometimes it won't say something the first time, but it will the second time.

This applies to probably all of the voices and languages. I even had the same problem when going on the website and trying it out.

This project is a good example: https://scratch.mit.edu/projects/370757928/

Edit: I think it could be a slightly different cause than this. What I described above is my theory on the cause.

It also happens to me! There are several bugs going on scratch!
I'm a proper scratcher but i cannot use cloud variables!