I used HRTF's all the time on the Studer D950S and vista consoles, but -and this is important- he's just starting with 2-channel stereo, and delivering to a 2-channel receiving device (the human head).
That means that the only 'processing' info from which anything can be derived is from comparison between the two channels.
HRTF in multitrack mix applications is one thing (and a thing of beauty, at that) but not really applicable to or in any way flexible for stereo-source-only, other than accommodating (extremely fixed) room circumstance. -Since he's obviously gone to great lengths to control the room acoustics (not feasible in real life) and the listener's position (again, not typically feasible unless you're something like an F-16 pilot) then I agree that this is all wankery.
I also agree that the reporter didn't sound able to think their way through what they were reporting on.
This IS transaural processing, it IS a parlour trick, and it's not very useful in the real world.
Also, let's not overlook that since different techniques are used to assemble stereo recordings, (spaced-pair, coincident mic placement, Pan-pot-multichannel, etc) then the inter-channel differences (timing dominant in spaced pair, amplitude-exclusive in coincident and panpot recordings) dictate that there is NO way to generate reliable "3D" from stereo. No more than you can generate reliable stereo from mono.
Nor -incidentally- is this Three dimensions. -This is a peeve of mine, sorry in advance for the rant.
A theoretical 'point' has no dimensions. A 'line' has one dimension. A 'Square' has two dimensions, and a 'cube' has three.
Similarly, Mono is non-dimensional because it resembles a point.. Stereo is actually ONE dimensional, because it resembles a line, where sounds can be placed along a single dimensional range between the two speakers. Quad/surround is TWO dimensional, because there is an added axis of control, as illustrated by the 'square' panning grid which you often see manufacturers use. "Three-Dee" requires an additional VERTICAL component for spatial 3D.
I really wish people would use the terminology correctly. Only things like Periphonic (a type of Ambisonic playback) and true Imax (where the sixth full-range channel is the 'voice-of-god' overhead speaker) can really be called "three-dimensional".
Anything else is marketing hyperbole, and utterly inaccurate. deriving a stable, position-and-reflection-tolerant controllable delivery from two speakers is bovine excrement. Nothing else.
BOSE advertise 'surround from two speakers'. That's a similarly fertilizer-laden claim.
You CAN perform some parlour tricks with two-channel stereo source, and DSP can be used to 'steer' between channels...
But then delivering through two speakers re-limits what you can do, and if you split it through MORE channels, -while you overcome some of the delivery/reception difficulties- you cannot be assured that your decoding is accurate, because of the randomized possible stereo picture assembly techniques.
-What of the Decca Tree now?
Keith