It may be difficult to pull off, but I think the place to do a null test like this is at the speaker level, recording the speaker playing a specific track with one cable hooked up, and then again with another cable hooked up, or maybe a different dac or whatever. This would test the cable's interaction with the complete system. As long as we were careful not to accidentally nudge the microphones or speaker position, I'll bet we could tell within 0.1 decibel if there were any actual spectral shift in the output.
If it's less than that, at least it gives us an idea of what level of subtlety we are dealing with. Even if there's no measurable spectral shift but a slight change in overall gain that could be telling.