Hi @badger101 , I think what you are seeing is basically a result of wrong character encoding.
The way I tackled these kinds of issues in the past was to check their hex values in order to identify them, and the remove them.
Alternatively, you can do the other way around, which is to instead identify what you want to keep (which @mlauber71 has already suggested and that you have implemented) and remove the rest.
If you are interested in looking at some sample threads that used the hex values, here are some: