As Mark often tells forum members, "You are always welcome to visit and see for yourself". Maybe you can invite him over and conduct ABX for him...
Business is business.
It's the playbook they (Jakob(#), jam & ...etc.) all use as if they get together regularly to share and coordinate strategies.
The attached chart that shows the limitations of several types of tests contains several blaring inaccuracies. In ABX, it says limitations are "No guidance over an attribute to focus on", which is quite inaccurate, both by your own statements and by those contained in the Clark paper. In fact, you can, and should, train listeners in recognizing attributes. The ABX is then assumed to be "less sensitive", where it is actually not unless done with too frew trials, testers, and lack of training. Stating a limitation is "relies on assesors memory" is just silly. The comparisons are instant, and while auditory memory is invovled, the testor may evaluate each choice for as long as he wishes.
Double blind testing is the gold standard in many types of research, including drug and medical testing.
I suspect that would define a futile waste of time. It's a belief system that isn't easily changed by someone not desiring to change. Lets also understand that a good ABX test is non-trivial, requires careful setup invovling precision level matching, and to produce significant data, takes time, and ideally a group of testers. It's frankly a lot of work to go through just to change a mind. Not worth my time.As Mark often tells forum members, "You are always welcome to visit and see for yourself". Maybe you can invite him over and conduct ABX for him...
@jaddie, Thank you for the last two posts above. Written as they are I am more on the side of agreeing with you.
If we can keep the conversation cooled down some there is something I would like to ask your opinion on. In another thread at: https://www.diyaudio.com/community/...-can-it-make-a-difference.384031/post-6972024 ...Post #160, Lars Risbo, PhD, of Purifi, and previously with Texas Instruments commented in his second paragraph about audibility.
Similarly in an ESS presentation document they talk about some of what it is they think audiophiles hear that maybe other people don't. The document can be viewed at: https://www.yumpu.com/en/document/read/23182504/noise-shaping-sigma-delta-dacs-ess-technology-inc
Starting around page 28 of the card deck they begin talking about one thing that some people can hear. They talk about some additional problems starting around page 39.
In both of the above cases what is being discussed looks to me like its (1) things some people may hear and others not hear, and (2) the things that may or may not be heard don't seem to be primarily a function of FR or of audibility thresholds. Rather, the primary factor appears to be something more along the lines of how individual brains separate out signal from noise (where 'noise' is being used here in the broad sense of consisting of any unwanted signal).
An example of such a separation task (maybe a kind of wetware filter function) might be understanding what your friend across the dinner table in a loud restaurant is saying. Obviously FR and hearing thresholds are factors too, but they don't seem to be the main factor involved in extracting the speech signals of this case from the unwanted noise.
Then my question would be to the effect that if we assume for moment that the described effects truly occur, what does that have to say about limits of audibility established by ABX testing when only fairly small fractions of he human population are used as test subjects. How would a researcher discover what the development team at Purify might be able to hear that most people don't hear? How would a researcher know about the ESS executives that have been trained to hear something most other people don't hear?
The reason I'm asking is that I am beginning to suspect that under favorable conditions, say, low stress and use of a familiar reproduction system, some audiophiles may have brains that have learned to perform signal extraction tasks (of low level distortions and or noises) that most brains have not learned. If true, maybe that could account for some of unexplained claims, counter claims, arguments, anger, rants, etc. that have been occurring in audio forums for decades?
Another thought along similar lines is that if something called 'expectation bias' occurs in hi-fi listening, how to we know its not that of people expecting to hear only transparent reproduction? How do we know if for them only pure music is the signal, and little reproduction imperfections are the noise to be separated out?
Having written that those last couple of paragraphs I expect to be attacked for having the gumption to question some other people's fixed beliefs. As you pointed out already its not worth the time and effort to try to prove something to a few very vocal people who don't want change. You're not the first person to make that particular observation BTW. I've said it before too.
If we can keep the conversation cooled down some there is something I would like to ask your opinion on. In another thread at: https://www.diyaudio.com/community/...-can-it-make-a-difference.384031/post-6972024 ...Post #160, Lars Risbo, PhD, of Purifi, and previously with Texas Instruments commented in his second paragraph about audibility.
Similarly in an ESS presentation document they talk about some of what it is they think audiophiles hear that maybe other people don't. The document can be viewed at: https://www.yumpu.com/en/document/read/23182504/noise-shaping-sigma-delta-dacs-ess-technology-inc
Starting around page 28 of the card deck they begin talking about one thing that some people can hear. They talk about some additional problems starting around page 39.
In both of the above cases what is being discussed looks to me like its (1) things some people may hear and others not hear, and (2) the things that may or may not be heard don't seem to be primarily a function of FR or of audibility thresholds. Rather, the primary factor appears to be something more along the lines of how individual brains separate out signal from noise (where 'noise' is being used here in the broad sense of consisting of any unwanted signal).
An example of such a separation task (maybe a kind of wetware filter function) might be understanding what your friend across the dinner table in a loud restaurant is saying. Obviously FR and hearing thresholds are factors too, but they don't seem to be the main factor involved in extracting the speech signals of this case from the unwanted noise.
Then my question would be to the effect that if we assume for moment that the described effects truly occur, what does that have to say about limits of audibility established by ABX testing when only fairly small fractions of he human population are used as test subjects. How would a researcher discover what the development team at Purify might be able to hear that most people don't hear? How would a researcher know about the ESS executives that have been trained to hear something most other people don't hear?
The reason I'm asking is that I am beginning to suspect that under favorable conditions, say, low stress and use of a familiar reproduction system, some audiophiles may have brains that have learned to perform signal extraction tasks (of low level distortions and or noises) that most brains have not learned. If true, maybe that could account for some of unexplained claims, counter claims, arguments, anger, rants, etc. that have been occurring in audio forums for decades?
Another thought along similar lines is that if something called 'expectation bias' occurs in hi-fi listening, how to we know its not that of people expecting to hear only transparent reproduction? How do we know if for them only pure music is the signal, and little reproduction imperfections are the noise to be separated out?
Having written that those last couple of paragraphs I expect to be attacked for having the gumption to question some other people's fixed beliefs. As you pointed out already its not worth the time and effort to try to prove something to a few very vocal people who don't want change. You're not the first person to make that particular observation BTW. I've said it before too.
Last edited:
Looks certain way to you and seem to be? Do you have anything other than speculations? You know, like evidence.
Not beliefs but observations. Numbers of DBT on audio cables, DACs, amps have been conducted and the results reported. They proved the boutique audio business wrong. OTOH, those boutique audio business' haven't reported DBT supporting their products' superior audible sound quality. Why? It's either they haven't done such test or have done it but the results aren't in their favor so they swept it under the rug.
He's reporting their findings. Nothing to disagree with. He talks about "training" for better difference detection, and I don't disagree, never have.
I don't have time to go through that right now...work demands. Perhaps later.
Agreed with (1), never been a dispute. I'm not sure about (2) being a function of only FR, though FR differentials are audible as a function of the total area of the response curve affected. It's bandwidth * gain = audibility. And "audibility threasholds", as related to nonlinear distortion is the holy grail we don't quite have quantified, though we do understand audibility is a function of many factors including signal type and harmonic complexity, and masking by other signals, etc. Complicated to quantify, but that doesn't mean we don't know anything.
More like individual ability/aptitude along with training. Just like some people can't hear pitch well, and others have perfect absolute pitch. Abilities are varied, but abilities can be improved by training. It's entirely valid to train testers.
Actually the "cocktail party" hearing ability is far more complex thant FR and thresholds. It's very much a spatial hearing and a particular type of perception that's wrapped in congnition. Even those with impaired hearing can retain, or loose, the cocktail party effect somewhat independently of hearing loss.
That's a very different research project than attempting to determine a particular mechanisms audibility. And, you have to pick your "small fractions" well. One test I did invovled a wide cross section of economic strata, listening abilities, age and sex. It's invovled, but not impossible. You're always going to be careful about extrapolating a generality.
That would be part of the test background. If you don't ask the right questions, and something like training background would be pretty important. It's also something you can pre-evaluate with another kind of test.
Yes that's true, but with any group you have to also consider bias, which is a huge factor in what is perceived. The placebo effect is well known, and proven to swamp out actual observations when the differentials are small.
Yes, bias is powerful.
I'm very cautious about "pure" as it relates to a signal though. It's never really anything like "pure" when it comes to a recording, but we're going for "unmodified" relative to the best we can get of the storage medium. But pure...probably not the best word.
As to someone separating imperfections from "noise", well, just ask them. In the end, music enjoyment is affected by a lot of things, or not. All systems are big signal modifiers, just some less big. Some get full enjoyment from a rather distorted system.
Actually I hope the dialog is helpful and continues.
Sagen, there is no "signal path". It's a current loop. Resistors: if changing a resistor for a different type of the same value objectivly changes the sound get your scope and check that node for instabilities. Fix that. You will enjoy richer tone colours and all of that with the cheapest of resistors you can find.
resistors are not created equal - some are more nonlinear than others. We had a whole production batch of Eigentakt amps failing end of line test because the wrong type of resistor was used in one single location on the PCB
That would be true of every component, passive or active. The real question is one of degree. That makes the statement above meaningless without numbers and specifics.
Misapplication of a component type often results in fireworks. Was this failure specific to resistor nonlineariety or just a misapplication?
Meaning: was it "wrong" because it was, say, a 1/8 watt type used when 2 Watt was needed (i.e. a Technical difference) or it was "wrong" because it was the wrong colour or built on the "wrong" phase of the Moon?
If the former, it can very well make an end of line TEST fail; it will most probably involve test instruments, measurements, even audition tests detecting audible differences.
If the latter, meaning they "failed" after somebody found "oh my God, we used the wrong COLOUR/BRAND one!!!!" , means it can´t be taken seriously.
If the former, it can very well make an end of line TEST fail; it will most probably involve test instruments, measurements, even audition tests detecting audible differences.
If the latter, meaning they "failed" after somebody found "oh my God, we used the wrong COLOUR/BRAND one!!!!" , means it can´t be taken seriously.
When I was 16, I worked at an electronics store. The Bose stuff was far away from all of the other speakers, so you'd magically think the AM5 was worth the price because you forgot what you just heard already 🙂
Please elaborate, did a letter get switched around and thick-films instead of thin-films were used? I don't think that is the point here, most interested people know about the basic differences in SMD resistor technology.
I do not cheat myself or are imbued of supposed "Audio Superiority" nor golden ears or similar nonsense, rather a quite practical feet on the ground problem solver, so knowing my own limitations, for speaker comparison,both for own and Customer benefit, built a 2 speaker cabinet with a footswitch controlled relay inside, so I can instantly switch back and forth between 2 speakers.
Only way to catch/show even subtle differences, such as a couple more/less wire turns in a voice coil, different former materials, adhesives (both type and quantity), even different dustcaps,different cone batches, fresh/softened (a.k.a. "burn in") speakers, etc.
Different time tests become unreliable by comparison, unless differences are gross.
Thats because you actually have electronic experience/knowledge not just over active imagination.
Maybe maybe maybe. It can all be measured and nothing here is new. Theres very few instances where resistors (and zero when there specs are the same) make an audible difference, and these extreme examples are well known to competent designers. Blind test or your just another mouth piece.
Hi, everyone. Lets try this dilema then. I have only two set of resistors, cant by other. So, put the type of resistor that is the more stabil one and clean sounding as first resistor on dac board where I2S come in, or as last before dac chip? Im talking about the 22 Ohm resistors on the board,. I try to keep this easy, but I see this post taking of now. AND, that is ok, alot to learn here for us all I think ( : Frank
Last edited:
Well he plausibly explained how two different resistors could sound different somehow in a specific application. But the general claim I really want explained is how said resistors have the predictable effect wherever they are used in a signal path. Of all the various points in circuits at different signal levels, with different currents, in filters, in attenuators, in tuned circuits , in your mothers cole slaw, they always make things warmer or more defined or less salty.