Nice! The german version breaks Umlauts though. Apparently the preprocessing converts e.g. "ä" to "ae", "ö" to "oe" and so on and the text2voice model subsequently pronounces them as e.g. "a-e", instead of what "ä" would actually sound.
I studied German at school but not enough.
Vielen Dank für Ihren Kommentar, er hilft mir, das Tool zu verbessern :)