I get duplicates. My job is to deal with that, on a scale of huge corporate datawarehouses, using SQL Server, and to carefully create a complex query to pinpoint the root cause.
Having to remove dupes manually is painful, once the count is above 200. Just order the dupes by name, or bitrate or date_added if you prefer, and cream them off into the trash. I wish there was an option for that in Automator; I have to say it's dismal!
I considered importing and analysing the iTunes XML file into SQL Server, which would take 0.001 seconds to show a report of the top... 1,000 dupes, and the count per tune, but then no real way to go and delete these automatically, going by date_added or choose the lowest or highest bitrate, resulting in a copy per library. Oh, Apple