It's curious that you emphasize the shoulders - don't tell Tchad Blake his expensive,
shoulder-less Neumann KU100 isn't binaural. The earlier KU 80/81 had no shoulders either. Nor the head that was available for the Sennheiser MK2002. The shoulders become significant for scientific analysis (Aachen head, Head Acoustics, etc.), much less so for music recording.
For recording music, a vaguely head sized and shaped obstruction between the mics is all that's needed. Try it for yourself - with and without pinnae; with and without shoulders -
for music in a reverberent environment, the difference is pretty insignificant - except that w/o pinnae, the frequency response is much smoother, so doesn't require complex EQ.
Anyone who's experimented with eyeglasses-mounted mics (as I have) has heard the incredible all-around-you realism - the hallmark of binaural - that they produce. These place the mics well away from (not inside) the pinnae, thus avoiding their frequency-distorting effects. The shadowing effect of the head is a good 95% of binaural realism. Pinnae only affect frequencies above about 3kHz; localization is mostly below that. Note also, that Head Acoustics'
BHS II binaural recording headset (shown at the bottom) places the mics entirely
outside the pinnae.
One can divide binaural phenomena into micro and macro categories - the brow/nose, pinnae, and shoulders are well withing the micro; for music recording, dispensing with them is largely inaudible - the overall head size and shape (and placement of the mics on it) is what matters most. The other thing I noticed in my experiments, is that the micro effects of the brow/nose, pinnae, and shoulders becomes less and less noticeable as distance from source increases; which is why they become much less significant for music recording at classical distances (great for those barber shop demos; less so in the concert hall).
I trust my ears (no pun intended) - I've been experimenting with this stuff since the '80s, with both nature recordings and music.
One can certainly argue that technically, the term 'binaural' should be reserved for methods/devices that include pinnae, but for the practical purposes of music recording, I'm quite comfortable referring to a pinnae-less head as binaural, since their
spacial qualities are largely indistinguishable in that context. Quasi-binaural
would be a more accurate term.
Bruce Bartlett's book is a treasure trove of info on binaural, quasi-binaural and transaural techniques (1st edition; haven't seen the 2nd).
https://books.google.com/books/about...d=_c48BPY_lIQC