Problem

We had problems with utf-16 TMX files before but they were solved by change of xml parser and the way process is setup.

Now we introduced additional pre-processor that looks for invalid xml characters in TMX as we now that some of them may actually have them one way or another.

But when implementing this we created a check for utf-8 characters and forgot already that files may come in different encoding as well.

Solution

Convert files to utf-8 before processing them.