Sadly, when my wife came home from work, even she was able to beat my score...
But here's the key: I've now refined the note authoring such that it reflects how I *actually* sing it. While the missus is very familiar with the song (her favorite of mine), she was singing it how it was on the *screen*.
So I've put in the proper bends, falls, and various inflection so it's a better match to my actual performance, and can now score in the 5* 98% range. Whew! Confidence restored!
Also boosts my authoring confidence: I'm getting very familiar with the process so it's second nature. It helps that Reaper instantly allows you to bind "actions" to any key-combo: that's one thing that I wish SONAR did, so it looks like I'm going to be using Reaper as my authoring mainstay for everything after exporting the stem mixdown from SONAR. Right tool for the job, as it were.
I'll probably do a brain-dump post on vocal authoring once I think I have it (mostly) under control. It sort of makes me wonder if the "live" wiki doco site has a lot of this stuff already, though: will it be redundant when we finally hit the "real" open beta?
Thanks also to others for the general welcomes and messages of encouragement.