I may be deceived by the video. It appears D&D may use a trick, which is record the performer in an anechoic chamber. Then, if they do live vs. playback in a dead, absorbent room, they stand a good chance of sounding close or even indistinguishable.
Acoustic Research (AR) was famous many decades ago for a similar demo in a room in Grand Central Station using a string quartet that would silently "bow synch" passages to the recording or play live. The taped recording had been made in the anechoic deadness of an outdoor setting, and the playback room was fairly dead. People were impressed after they found out they were hearing a recording vs. live. But, the visual cue of the "bow syncing" probably helped fool the minds of listeners.
I suspect that as long as there is a reasonable acoustic match between recording and playback venues, a lot of today's better speakers could pass this type of test even in blind stereo listening. But, at the same time, stereo just does not reproduce an adequate replica of what we hear in a live concert acoustic, as with classical music, at least not for me.