[rrrpiano]

Thanks for the ideas everyone. As Sciurus says, all (present) methods require an index. I guess I was hoping that there was a "magical tool" that could do this without an index...

I'm no software guru, but I could see how a tool could input a large pdf songbook, reading it page by page. If a page had >40 point text (for example) at the top of the page (the song's title/name), this would be marked as a "first page" (and this large text would also be used as the split-out pdf name). The tool would continue reading the next pages, until it hit the next "first page", or the end of the file. Just dreaming ;-)