A measurement, if carried out seriously and professionally will objectively describe a certain
characteristic of the device under test (
DUT) under defined and fully specified preconditions. It is valid if it can be reproduced by others under the same boundary conditions. Nowadays, many very reliable measurements can be taken by hobbyist with limited, but still noteworthy effort.
The problem is
not that music was different from measuring signals in some mysterious way. The problem is
not with measurements not telling the truth.
The problems (and there are many) are (mostly) still the same as they used to be from the early days of electroacoustics.
- Know and understand the influence of your boundary conditions. What effect do deviations have on the result?
- Know and understand what linearity means and don't fall for conclusions or interpretations that are only valid for linear systems when this linearity hasn't been proven, yet.
- Know and understand that one single number characteristic of a device will never describe its behaviour completely.
- Know and understand that improvements regarding one characteristic might negatively influence another one and it might not be clear at all which one is more meaningful to the result.
I could probably go on for hours. And fhi is where all the myths kick in: When single measurements seem to not align with what somebody likes or doesn't like to hear.
In the end for us as consumers, it makes little sense to buy a device you know you don't like how it sounds (unless you have that nagging fear in a far away corner of your brain that you might simply be wrong). But it also makes no sense to deliberately buy stuff with known bad measurements and try to find "synergies" between technically bad components.