It is possible to measure distortion somewhat below the ambient noise floor, though. And, at least in the case of transducers, it’s possible (and preferable for most purposes IMO) to measure at distances below 1m, which increases the SNR.
Having said that, I can’t speak for
@tuga, but when I said “low-level”, I had in mind levels more in the 65-85dB range, which represent the average SPL for moderate-level listening, or peak SPL for quiet listening. Basically anything below the more commonly measured 90 to 100dB/1m.
I agree there’s generally not much point measuring at such low levels that it’s difficult to separate distortion from noise: if our instruments struggle to do this, our ears certainly will, too.