...direct physical contact of the source with the skull that is.
Al,is spot on, in regards this
It's currently believed it requires direct contact
I have only seen a few publications suggesting thru air perception
One suggested we can pick if the ultrasonics have been removed , but cannot hear the ultrasonics on their own
However as I have stated before the main issue for ultrasonic content in music reproduction are 2
1. It's is important in the time domain
2. Attempt to remove ultrasonic content in digital recordings creates other problems downstream if the sampling rate is insufficient to be above the "natural" musical pass band
Many have commented that most older microphones only spec to 20khz
Also the energy content falls with increasing frequency
However, because of the high dynamic range of cd, even if the musical spectra in the ultrasonic region is 40 DB down, it will be folded back down into the auditory band without filtering
And much of this will go into the 0-7 kHz area where the ear is most sensitive
Brick wall filtering will cause preringing, but block the fold back
Please note the patent on the aphodising filter , avoids preringing but is specified to allow 32db of breakthrough in the 0-7 kHz band
To my mind I am increasingly favourable to the idea of MQA as a technology
But ultimately the market place and people's ears and wallets will decide