Even though I was under the assumption that the umlaut was part of the Unicode character set, I never, ever use terrestrial characters to avoid the classic character encoding problems.
So Frieder to answer your question, no I did not use the umlaut in my audio nor projects.
But to confirm your bug, I copied some umlaut characters from the web, for my US keyboard does not create them and used the ö, Ä, é, and ß in some project and audio names.
After 2 recording takes that had some trim and normalise actions, I used the undo button, and presto; All audio tracks are empty!
I’m glad you guys discovered this bug, for you are getting close to solve the problem.
Keep up the good work!