Aside from "stadium evacuation requirements" and the like, most people's lives do not depend on "perfect sound". Perfect sanitation during chest surgery, though, is a thing. Apple, orange.
Seriously, because the overtones above 10K (I'm a bit better off than you, and a year older, but hearing loss is part of life in the way-too-loud modern world) are not the primary stuff that defines what music is. In many recordings, nothing over 10kHz happens to a level where it's actually audible (i.e. above absolute threshold) for a person with normal hearing. A close-miked instrument with lots of high frequencies, of course, does have energy up there, but you still know that was a cymbal, and the crash is what makes the music, not the tippy-top of the spectrum, UNLESS you're trying for something specific.
Why not? Engineers can do this too. Anyone can do this. Listen to what you PREFER.
Well, there have been some recent arguments about that, but I'd suggest, in the sense you want a meta-discussion, that isn't it really all about what YOU want to hear? Face it, Zeppelin on an 8-track is still cool with the spring reverb and the speakers cut into the back panel of the 1966 Mustang, and no, that's not "acccurate" at all.
Does it matter? Seriously. A good USB stick is pretty good, too, as long as you can handle the ground noise.
Depends if you're out of beer.
If you want light, turn it on. If you want dark in the day, pull the shades. It's about PREFERENCE.
<insert reference to Queen song about Sunday Afternoon or maybe Moody Blues song about "Tuesday">
Rooms outweigh everything. "room correction" can handle some stuff, but a bad room is a bad room, oom, oom, oom, and once that energy is stored in the room, it's there until it runs out. Speaker placement likewise, but "room correction" to some extent is "speaker correction" and it can handle some amount of weirdness in placement. Not too much, though.
There are a bunch of electronic categories where the engineering issues have now been "solved" for years... in those areas of performance, audible differences between components are either intentionally engineered, or a sign of very poor design/manufacturing.
Thank you so much...much better put than my version...
At least it was in his roundup article about thresholds in that 2011 article you linked, but I think over time he may have updated his estimates of when crosstalk really becomes inaudible, as one can see over his multiple device-specific measurement and review articles he keeps calling results around -60 dB "good" or "very good", but only says "excellent" once he sees -80 or so. So in my mind the NwAvGuy threshold for crosstalk is really somewhere at -80 dBr.
Why think about this stuff at all when reliable brands far surpass these thresholds?
Because sometimes "upgrading" has an underlying expectation of audibility, not engineering excellence.
I can try to translate what I already said into some other language if English poses problems for you. Try to read again, especially the phrase about dBu.
Have ever looked at crosstalk measurements coming out of record players? -30dB or worse is normal. Crosstalk is just not very audible by itself. Think about how much crosstalk occurs acoustically with speakers or microphones. Think about how much is necessary, with plug-ins, to make a difference in headphone presentation (you'll easily hit -10dB).
This is just a cheap shot trying to "send to bed" anyone asking for better published specs from manufacturers and/or better measurements from reviewers. I can't measure these things personally in a way that will tell me what I would hear in every life situation, in every biological state, with every song I will ever listen to, with every speaker and pair of headphones, and in the case of DACs with every amplifier I will ever hook them to. That's why general limits of human hearing exist (i.e. this topic and the thresholds in the first post), and that's why generally informative published measurements exist. So that I don't have to worry about redoing all the damned research in my room at home.
The thresholds at the beginning of this thread are very, very generous from a perceptual perspective.
Weird choices of test signals, this doesn't look like a study seriously looking for the absolute lowest detectable level of crosstalk. The pure tones were not close to the frequencies of maximum human sensitivity, broadband noise seems mostly useless, as for the music samples... how are we supposed to generalize from these 3 songs in 1985 to all songs in existence? Since minimalistic passages (e.g. a single instrument + lots of silence) exist in music, and because a huge variety of synthetic instrumentation exists today with far less harmonic content than natural music (thus far less masking power?), this kind of experiment - if it was seeking the absolute limit of audibility - should be done with the most shrill and easily audible distorted signal played through one channel (with that side's earcup or earbud kept away from the listener, beyond physical audibility) and listened for only in the other channel.
Already addressed above - you don't know how much or how little masking there will be in every musical passage I will ever listen to. I won't waste my time discussing masking in the context of absolute audibility thresholds. One could just as well find that some sparse musical content could enhance the audibility of the crosstalk component via intermodulation.
Nothing to do with the crosstalk discussion, plus this test contains several errors that make its results unreliable: selected tones and IMD products very far away from the spectral region most people are maximally sensitive to; not allowing the test to go further down than -69 dB when the aggregate results plot shows a number of people suddenly gather at that last bar and would've probably been able to keep going down and made the distribution look more realistic, with a longer tail tapering off to who knows what ultimate value; also the test gets arbitrarily cut short after you make 3 direction-changing mistakes even if the first 2 were just "hiccups" along the way down and you could continue getting better and better levels after recovering from them, but at the 3rd mistake you are suddenly judged incapable to recover and keep going down in just the same way as you did 2 times before. Can't trust even the central peak bar in the population bar chart with this arbitrary cutoff nonsense in there.
I think he rates this based on what is achievable from the perspective of solid engineering, not audibility, or he might extrapolate from more straightforward THD or SNR thresholds. Either way, not really a problem.
This?
It's easy to cheat these tone tests on both sides. The several hundred who score -69dB, >90% are suspect. Here is what -69dB looks like to a cheater:
). Here is an answer I gave to one member who used tools to assist his perfect score:
www.audiosciencereview.com
We are also about 10dB less sensitive to distortion in music than to pure tones.
First idea was we need crosstalk measurements for every source device, not always provided today, and even when they are it's not with clear test conditions wrt. load and signal level.
Did you read what Amir wrote about what it would take to detect the differences 120 dB down from the peak signal? I recall him describing the special training he received and the strategies he had to undertake to detect distortions at -115 dB, and it requires severely gain-riding the tails of reverberation and such like. And that was to be able to detect a difference sufficiently reliably to be statistically significant, meaning just barely.
