I completely concur.
This is something that is a must.
Will be very happy to see it implemented.
At 95% sensitivity I still get hundreds of duplicates, and most of them are not.
Perhaps there is something wrong with the way I use it, or the algorithm gets confused.
I suspect when your collection us under 1000 songs this is a non-issue, but with libraries like mine 13000+ (i am sure others have way bigger collections) this is a major problemo.
Seb