Discussion in 'SBAF Blogs' started by purr1n, Dec 26, 2018.
Time is money!
The limits of audibility of various distortions have been characterized by scientific experiments. I think it's fair to say that a piece of equipment that contributes -130 dB THD contributes an inaudible amount of distortion (based on scientific experiments that reveal the audible threshold of THD is about -80 dB in the best cases). Similarly, if the noise floor of an amp would result in an SPL well below 0 dB(A), I would argue that the noise of the amp is inaudible.
Now, you can argue that your ears are much more sensitive than average. I have no data to dispute that. However, just as ~85% of drivers rate themselves as above average drivers, I suspect you could be dealing with the Better-Than-Average Effect (which is also backed by science).
Some just flat-out reject science. Possibly because science doesn't support their preconceived notions of how the world works. That's very in vogue these days. I'm not one of those people. While the scientific method has its flaws, I fundamentally believe its self-correcting nature will ultimately lead science to provide the best description of how the world works. You're obviously free to disagree.
I have never claimed that any of my products are be-all end-all, though some of my customers have described them that way. I make the measurements available for all to see. At least that way you know what you're buying. Should you put more weight on reviews by "trusted ears", you can always request the reviewer to review my products. One of my HP-1 amps also made the rounds here as a loaner amp. You can sign up for such a tour and form your own opinion. To me, this is an honest approach.
Ok, now that you've stated the numbers, could you explain what advantage your box has over a >80dB box? Or would it be fair to say any stated improvement claimed by users of your box compared to the >80dB box is placebo?
Have never implied any of this. Are you actually reading what I post or just running your "response_script.exe"?
If you only look at that one measurement, then yes. However, as I've been saying many times over, I rely on many measurements (plural) not just one.
You can look at hifi gear from the 1970s to see what happens if you lock in on one measurement exclusively. The equipment of that time showed excellent THD, but sounded like crap due to the slewing-induced distortion (SID). We now know what caused that and have changed our amplifier designs accordingly. That's why you saw measurements of transient intermodulation (TIM) distortion pop up in the 80s and go away as amp designers updated their designs. These days, you hardly ever see TIM specified. It's a solved problem.
Yes. I do read what you post. Do you read what you post?
All you have done is obfuscate by throwing out some other distortion tests that are similarly inaudible by your own admission. Since these are all solved problems I'll ask again, what does your design offer that a cheapo off the shelf box doesn't?
Or are we to take all of that to mean that your amp exceeds in some new distortion spec that only you are privy to?
On digital amps and Hypex:
Hypex does measure much better than other class D amps. So it's fair to say that they have done something different or kind of new.
I am curious about the Crown CDi series, but they are way more expensive than the XLS. I actually asked the engineer here (not the sound engineer, but the hardware engineer - the TOTL Crown Macro-Tech i Series amps are used for the L, LM, M, RM, and R) for the marquee stages. I know that the smaller mix stages have the cheapo XLS. The I series uses Crown's "class I", which is basically class D.
Back from derail:
My blog post was specific to SINAD or THD+N and really not an article on subjectivism vs. objectivism. My premise was that once SINAD is past a certain point (70-80db) it had no bearing on perceived sound quality.
Basically, @trung225's post above is a good additional commentary.
On blind tests:
Absolutely no reason to do them other than for curiosity and to demonstrate to oneself that the naysayers are wrong. As @OJneg said, the naysayers will always move the goal posts.
However, I am absolutely satisfied that I can tell the difference in triple-blind level matched tests where I feel such tests are unnecessary. I still think they are fun though, especially when two pieces of gear sound very close to each other.
On fraud for MOTs advertising -143db:
Maybe fraud is too strong of a word, but if someone does take a true scientific angle, there are so many other factors that will render such THD+N meaningless. Ambient SPL, transducers, human perception, etc. being the limiting factors. So maybe disingenuous? Now you guys know why I hate Benchmark and their whitepapers.
On the subjectivist to objectivists scale:
I don't see it like this. In very simplistic terms, I would be on the "B" on @OJneg's scale. However, it's never so simple. For DACs, where most of them measure so well past human and environmental limits, I'm an "A", subjectivist on the scale. Although if one focused on the filter transient response, I might be somewhere around "C".
So really, it's complex. It's which specific measurement and where the numbers or patterns lie with respect to human perception.
I generally don't see anyone doing science in regards to correlating measurements to what we hear. Measuring a SINAD 132db DAC and announcing it as "very highly recommended" and another SINAD 91db DAC and announcing it as "not recommended", or even putting SINAD of gear on scale... that is most certainly not science. That's religion. Unless one listened to recordings with peaks at 132dbSPL + 35dbSPL ambient.
On ABX testing or sighted AB testing being hard work.
That's the reason why a pile of gear that I need to "review" is sitting on top of the table. It's hard work and takes away time from just enjoying music.
I would guess that most folks on SBAF are a B, understanding that the measurement correlation can be stronger or weaker depending upon the type of DUT and the type of test employed.
From site analytics, I'd say that SBAF is somewhere in between A and B with a much stronger tendency toward A. Personally, I'd like it to shift a bit further to B because I hate when people say stuff like "Why didn't you guys criticize X headphones more?" when a glance at the FR would have told them this headphone or IEM might not have been for them.
Well, pushing the limits of how well a piece of gear will measure does seem fun and curious in a geeky way, and isn't dishonest. Also, there is a market for this. Numbers like this appeal to kids who grew up looking at FPS charts on AnandTech. Personally, I preferred HardOCP's FPS plots to see how even the FPS rates were and to note any serious FPS dips (which do disrupt gaming).
It's kind of a bad analogy, since some games when running at the highest resolutions with all the eye candy on can bring TOTL graphics cards to their knees; whereas any device with "bad" SINAD of 88db is more than enough for most gaming headphones. In other words, 44FPS to 87FPS average makes a difference, however, SINAD 132db to 88db, not really.
P.S. I think it's funny how ASR used to rely on the arbitrary "Amir-bits" based on linearity measurements (initially improperly done without narrow bandpass filter until Amir got schooled by Jude of all people - but it was probably AP telling Jude what to say) until @ultrabike brought up Analog Devices definition of SINAD.
I blame @ultrabike for the latest SINAD crazy, for giving Amir the idea.
As much as I measure I wish I could find the combination of metrics that might correlate to my perceptions (including my personal preference bias) such as given in the three example systems above.
Further, I'd like to see anyone with decent hearing and trained listening endure the "No Soul" stack for more than five minutes without ripping off the cans, maybe ten with a good analgesic.
To quote someone special:
What you hear is not the air pressure variation in itself
but what has drawn your attention
in the streams of superimposed air pressure variations
at your eardrums.
Words to live by. I firmly believe that brain looks and hears in contrasts and relative scales, and can quite happily level out any filters after some time to a high degree (if there aren't too obvious offenses, of course, in which case the product is shit anyway). Watching through a bit colored sunglasses and crystal clear ones after a while becomes quite the same thing, but those with filters might help with glare at some situations. Some people perfer one or the other, and this is what I believe is trying to be achieved with "by ear tuning". How's that for an analogy.
That said, I perfer crystal clear glasses myself, which I then put those ridiculous shaded clips on (EQ).
Please re-read Post #85.
There's no answer there.
Please re-read Post #86.
He overdesignes (imho) stuff to be as transparent as possible to the limit. I can respect this. It's nice knowing that what lies under the hood is proper, and some people will pay for this. Way better investment then say, unobtanium cables... but, to each his own...
Tilt your right speaker 5 degrees and you've just introduced audible coloration 10x that of what any amp can make. Or, move your earcup a bit around. Granted, it's not the same kind of offense, but still... And even so, I can walk around the room and move the earcups around and not hear anything changing that much. Lot easier to reach the "good enough" level of performance with amps, so I tend to agree with Ojneg here, even as I do deeply respect Tom's work.
Have a nice evening, all.
Tech Report stole my heart when they sort of pioneered using frame time measurements rather than focusing on FPS. Even the FPS over time stuff doesn't give enough of a granular look into performance.
I remember playing Fallout 3 and wondering why motion never seemed smooth. Little micro-stutters all the time. Drove me insane and gave me literal headaches.
Turned out it was because AMD cards, at the time, had awful frame time consistency. So, yeah, you can get 60 frames in a second, but it looks like garbage if they're coming it at inconsistent times.
Up until then, the usual response was to just insult people like myself and say we were imagining things.
And now GPUs manufacturers have not only dramatically improved frame time consistency, through hardware and software, but we have a whole market dedicated to async monitors to further improve on frame time consistency and deliver smooth gaming.
Seems like a trend everywhere, to say you think you notice something off, only for armchair and/or real engineers to tell you to pound sand for making shit up. Someone smarter than all of us puts the puzzle pieces together rather than crossing their arms and sticking to their guns, everyone is better off for it when new findings bring together subjectivity and objectivity, and the goal posts get moved for the next dumb collision.
Gaming: tHe HuMaN eYe CaNt SeE oVeR 60 fPs AnYwAy YoU aRe ImAgInInG tHiNgS pLeAsE sEe ThE aNaNdTeCh GrApH
Lighting: ThE cFl AnD lEd LiGhTs DoNt HuRt YoUr EyEs AnD cAuSe HeAdAcHe YoU jUsT nOcEbO yOuRsElF
Audio: cOgNiTiVe BiAs! HuMaNs CaNt HeAr DiFfErEnCeS aT tHaT lEvEl AnD yOu ShOuLd OnLy TrUsT a LiMiTeD sEt Of StAtIc MeAsUrEmEnTs
How come glorious planars with their better THD (omg the bass quality) and the like are worse at letting you hear differences in your whole audio chain than peasant dynamics that hit 10% THD at 30Hz?
Oh man I totally remember this.
I actually find this analogy to be quite good. The only thing that I'd add (that @Hands alluded to) is that your perceived FPS is also entirely dependent on monitor refresh rate (as well as pixel response time, but that's less relevant in this particular scenario).
The way I'm understanding things, the monitor refresh rate is like our "human perception," and the FPS is the "SINAD" (performance) of our system. So if the monitor has a 60 Hz refresh rate, a pc that can hit 130 FPS consistently is going to be perceptually the same as a system hits 80 FPS consistently (barring any frame time stutters).
Of course, I could also be (most definitely am) crazy.
Nothing is firmly crystal clear and eq can’t fix timbre. You can try to compensate but you ultimately cannot correct what has been done. You yourself believe that whatever you own or believe to be transparent truly is transparent. That is very self-centered. It’s the same fallacy as the objectivists believing that whatever they can afford themselves truly is transparent and does not have a sound.
It’s the same as the redditors shilling the Behringer interfaces now. The conversion is muddy and the Midas designed mic pres are dull garbage proving yet again that Behringer is trash compared to almost everything else. Yet to them? Transparent!
None of what Tom makes is perfectly transparent. Nothing is. His head amp is way less transparent of a pre and a headphone amplifier amplifier than products with higher noise and distortion measurements.
Tilting your speaker won’t change the timbre of the amplifier. Stop being evasive. Listening to something on axis instead of 10 degrees off that is powered by a 1980s Yamaha natural sound receiver won’t bring back the detail.
Separate names with a comma.