The point is that what we normally measure and how we hear are not necessarily the same things.
Usually when the sound changes, _something_ will be measurably different, but in many cases it won't be anything as simple as a 20 - 20kHz frequency sweep.
The MC cartridge loading example that I gave in my earlier post is an easy and common case of audible changes that aren't reflected in typical measurements. If you load a cartridge down more heavily, the absolute level of all frequencies will diminish, but the relative level of 10kHz as opposed to 1kHz will not. What will change is ultrasonic stuff above 70kHz, sometimes extending beyond 1Mhz. What appears to be going on is that the phono stage (not the cartridge) can misbehave due to the ultrasonic, high-amplitude signals, and the ear (mis)interprets the intermodulation artifacts as "brighter" or "darker" sound. Such intermodulation has no harmonic relationship to the original signal, and the ear finds such inharmonic distortion far more nasty (and therefore noticeable) than with harmonic distortion.
Tonearms and headshells (and cartridge bodies) affect the perceived tonal balance, but this kind of stuff appears as blips or steps in the mechanical impedance response. You won't see it appear directly in the frequency response.
To borrow someone else as an example, I know that Syntax hears the Kleos as having much less upper frequency energy than the Delos. But do these two cartridges measure differently at the top? Very little difference; more than enough to be swamped by changes in the operating temperatures. Does this OTOH mean that Syntax is imagining things? Hardly - the body structure of the Delos and Kleos are quite different, resulting in different mechanical characteristics, which can and do give a different sound. FWIW, I hear the Kleos as having a "wetter" top end than the Delos, so Syntax and I are definitely hearing the same things. Where we differ is in how we interpret what we hear.
In general, I find that frequency response correlates to what I hear more so with amplifiers than with transducers. With transducers I get better matching to what I hear by looking at things like the mechanical characteristics (resonances and impedances) of the body structure or mount, distortion, and transient characteristics.
The likely reason is that distortion means that new high-frequency components are being generated. 2nd-order distortion on a 5kHz signal results in 10kHz artifacts that were not present in the original recording, 3rd-order distortion creates previously non-existent 15kHz artifacts. "Frequency response", OTOH, means greater (or lesser) sensitivity to signals - that may or may not be in the recording. Even if a component has a rising top end, if the recording doesn't contain signals at those frequencies, you will not hear it.
But even with amplifiers, the correlation between measurement and ear isn't always obvious. Change the power rectifier diodes in an amplifier and the tonal balance may change, but you won't see it in the 20 - 20kHz response. You will see it if you use a wide-band spectrum analyzer on the power rails, as "hash" extending out to 50kHz - 100kHz.
FWIW, I also hear differences between film capacitors - teflon as opposed to polypropylene, or tin foil as opposed to metallized aluminum, although when you go to measure this stuff, it will typically measure as -120dB or even less. A good resistor will measure around 130dB down, while a metal-foil Vishay will measure between 150dB - 160dB down (the smaller the signal, the trickier it is to make valid measurements). Down at this level, things shouldn't really be audible. But they are (smile).
kind regards,
