Use Lingua::EN::NameCase.

You're not dealing with words, it appears you're dealing with names. Therefore, use the below example in order to take advantage of all the special cases that ucfirst lc will miss.

use Lingua::EN::NameCase qw( nc );

use strict;

while (<DATA>) {
    s/(?<=AU: )(.*)/nc $1/e;
    print;
}

__DATA__
AU: jane smith
AU: JOHN DOE
AU: keith miller
AU: PHILLIP LEIGH-WILLIAMS
AU: PAUL MCCARTNEY
AU: DOUGIE O'CALLAGHAN
AU: MARIE ST. JOHN
AU: henry viii
[download]

The above code runs everything after the AU tag through nc and outputs the result, like so:

AU: Jane Smith
AU: John Doe
AU: Keith Miller
AU: Phillip Leigh-Williams
AU: Paul McCartney
AU: Dougie O'Callaghan
AU: Marie St. John
AU: Henry VIII
[download]

- Miller