The snippet you show is encoded in UTF-8.

Next step: determine the encoding of the file in which umlauts display correctly on your terminal.

Or even better: configure a clean UTF-8 enivronment.

I suppose the confusion lies in => if I create the file, I get my Latin-1. If I didn't create the file, there is only ASCII.

I'm confused indeed. If you don't create a file, it doesn't exist, neither with ASCII nor with UTF-8.

Speaking of confusion, I think you try to achieve too much in one step. For example the title of your question metions HTML::Parser, which doesn't appear in the posting at all.

So, small steps:

• Make sure you know which encoding your terminal understands. There's no point in proceeding before you have done this step.
• Find out what encodings your source files are. Seems to be UTF-8.
• In your perl scripts, decode everything coming from the outside (except when a module does it for you), and encode everything. use utf8;, and write your program files in UTF-8.
• If something doesn't work, find out where you violate any of the points of the previous steps.