Personally, I want filenames to be mostly ASCII, so I wrote Text::CleanFragment to create URLs or filenames that don't contain accents or other stuff. Mostly, it forces the text to match  /^[-._A-Za-z0-9]*$/ , which I find pleasing and which doesn't really have problematic characters for filesystems or HTML.

The module differs from your approach by converting (say) Motörhead to Motorhead, instead of Mot rhead, , and by compressing multiple dashes into one.