Ohm Walsh Micro Talls: who's actually heard 'em?

Here you go, the original measurement of the Walsh driver...
http://s1248.photobucket.com/user/peterhorvath6233/media/image_zpsomfobhuk.jpeg.html?sort=4&o=8

As for roll off, did you measure it? Or was it just feeling that it was extended way past 16k? remember unless your'e under 30 the chance you can even hear 12k is slim. 

From Jan 17, 2008 Stereophile review of Walsh 5 speaker:

" I was sure that a 1/3-octave frequency sweep would reveal an aberrant frequency response, but I was wrong. In fact, from about 500Hz to 20kHz (which was as far as I measured), the Walsh 5 was almost ruler flat when measured on the supertweeter axis. There were the usual dips and peaks in the in-room bass response, and the bass was down 4dB at 25Hz—in complete accord with Ohm’s specifications. The source for the coloration remains a mystery, therefore.
Read more at http://www.stereophile.com/content/ohm-walsh-5-loudspeaker-page-2#BMwfiIR8TJXhr0v8.99"

aftewrwards (from OHM site) :

Stereophile Magazine called the Ohm Walsh 5s: "one of the best American box speakers made"

Years ago Stereophile Magazine reviewed the Ohm Walsh 5 speaker. Their conclusion back then was that current model had some unique qualities and was on the "verge" of greatness. 
In response to that review Ohm made some adjustments and thus a new "5s" version of the Walsh 5 speaker was born. Stereophile reviewed this new model and in conclusion they had this to say: "a Great American Speaker—one of the best American box speakers made"

" I was sure that a 1/3-octave frequency sweep would reveal an aberrant frequency response"

That doesn't sound like someone who does a lot of testing!
A 1/3 octave resolution is as corse as my software allows to go and I would never use that for serious testing. It's kind of using 10000 iso for 
portrait photography or gravel instead of sand for a foundation not to smart.
Look at the readout, it gives you a feeling for whats going on but thats it.
This readout was me watching TV, not testing.
http://s1248.photobucket.com/user/peterhorvath6233/media/image_zpstgppmvnp.png.html?o=0

There is a huge discrepancy. That is flat response measured and judged one of the best speakers of its time by a leading publication in the industry and yours is way out of the ballpark for reasons unknown. Yours do not look like anything I’ve ever seen measured from any decent pair of speakers working properly. Something is wrong. I doubt it will matter that much but try 1/3 octave on your gear maybe and then apples and apples comparison can at least loosely be made. If not comparable then I do not know what to tell you other than your measures and those from Stereophile  are radically different for reasons unknown.