I don't think a pipe organ is a particularly close approximation of a human voice, because there's no musically controlled dynamic expression within the note. B

I don't think a pipe organ is a particularly close approximation of a human voice, because there's no musically controlled dynamic expression within the note. By this I mean each pipe can only produce the sound of the note turning on, the sound of the note sustaining, and the sound of the note decaying, and these are substantially the same every time. (There are subtle differences, because the air flow depends on the state of other pipes playing at the same time, but this is not something musicians actively control, and is traditionally considered a defect to be minimized in high quality organs). It's possible to modify the sound with a tremulant or a swell box, but these are crude compared to what a skilled singer can do. Compare the expression you can get out of a saxophone by varying embouchure and breathing. A saxophone can produce a range of timbres, and the player can dynamically vary the timbre over time. It's mechanically similar to a reed pipe, but the player has far more control over it. And a flute is similar to a flue pipe, again with the player having more control over the sound. But a real pipe organ has some measurable advantages. First, the great multitude of independent sound sources. A real pipe organ has a spatial quality that's impossible to reproduce with stereo speakers. People say that the building is a part of the pipe organ, and that's IMO justified because all these sound sources interact with it produce an incredibly complex reverberation field. The sound you hear when you move your head sounds "big" in a way that would require a very complex speaker setup to reproduce. (Maybe high-order Ambisonics could do it.) Second, the non-existent distortion. Big pipe organs have pedal sections that produce very deep and loud bass. On a real pipe organ, this is a pure harmonic sound (like all wind instruments, bowed strings, and voice). But deep and loud bass is difficult to reproduce with speakers, so there will be substantial non-linearity. If you're reproducing only a single harmonic note, this will only result in harmonic distortion. Harmonic distortion is no big deal, because it's effectively the same thing as changing the voicing of the pipe. You can filter the recorded/synthesized sound to compensate. But if you're trying to play multiple notes at once with the same speaker, you'll get intermodulation distortion, which adds a "muddy" inharmonic sound you never get from real pipes. The only solution here is adding more speakers, especially for bass notes, but again you're increasing cost and complexity. I don't know how complex a sound system would need to be to reproduce a pipe organ such that a skilled listener couldn't distinguish it in an ABX test from the real thing. It would be very difficult to test, because the two would interact with each other acoustically if they were in the same building, and an ABX test requires rapid switching. But in any case, I think most people simulating pipe organs aren't even trying to pass this theoretical test. All this means there is still a place for traditional pipe organs, and if you've never heard one in person I recommend trying it.