Quote: Okay, so these headphone amplifiers are real-world systems and thus not 100% textbook LTI systems. We can even quantify how "off" it is by looking at symptoms like THD given a certain input signal, load, and output level. The less "off" it is in various senses, the more we expect the behavior to be like the ideal "wire with gain" or whatever you want to call it. If you hypothetically had two different "wire with gain" then they should sound identical, right? The signal being sent to the headphones would be exactly the same for either. So if we had two "wire with gain plus no more than a really small disturbance like 0.00000001% added in" then surely those would sound the same too. One question is then about the level of disturbance that is still so small it can't be detected. Intuitively this seems like this should depend on the listener, the kind of music, the listening volume, the headphones, the type of disturbance (for example, 2nd order harmonics are of course much harder to detect than something else), and more. There's probably not a one-size-fits-all perfectly exact answer, but some studies have established some kind of safe estimates, with a lower estimate of the threshold being safer. The other question is about how indicative the typical audio benchmarks are of real-world performance with music into headphones. Usually bench tests are with resistive loads using combinations of test tones. With a fairly low-output-impedance amp that's stable, performance into non-resistive loads like headphones should be pretty close to the performance into resistive loads. Music is just a combination of tones, but it will in general contain more of them at once than you'll see in a typical IMD test. Anyway, keep in mind again that the more ideal the response is, the less distortion we should expect of all different kinds. The THD and IMD tests can be thought of as some kind of predictors of behavior. If the values are high, then it's hard to say. If the values are low, then we probably expect performance to follow the ideal fairly tightly. Has anybody ever seen an amp that performs really well at IMD CCIF and SMPTE into different loads and output levels, have high distortion with any other kind of input signal? It doesn't seem likely that this should happen. Also a trip down to InnerFidelity will show that the difference between headphone drivers (left and right side), not to mention sonic differences from slight differences in headphone positioning, are all way way higher in terms of FR and THD and so on, than between headphone amps that are considered to be probably sufficiently close to "wire with gain." edit: on a side note, I'm waiting for headphone reviews where they go into the night and day differences between the left and right earcups. "When paired with my überAmp, the velvety midrange of the left earcup is unmatched, such as when retrieving the low-level details of Lilliputian feet scampering about in the morning dew. But when combined with the authority of the right earcup in resolving every dynamic, the result I feel is a sound stage that is engaging with some music, yet muddled with certain passages. Rather than placing you in the second or third row, you are unceremoniously dumped in the second-and-a-half row, hardly a comfortable experience."