Thank you for the reply. Yeah my Kindle (10th gen Paperwhite) sadly doesn't seem to be reading any PDF metadata, it just parses the filename and guesses which part is the author and which is the title, so filenames are important in this case.
Unfortunately using "save to disk" didn't avoid the conversion. I've just noticed that Calibre already converts the filenames when importing the file in the database, that seems to be where the latinizations are coming from. There's no way to avoid this?
I can open Kindle's folder and rename the file after sending it via Calibre and apparently Calibre will still recognize that it exists on the device, so it's not a big problem, but it would be nice to skip the manual renaming
