AS you think that ,you are the ones that think all sound the same ..I suggest a garden forum..yep I can btw as a lot do here.
You have accept that caps have different TDH profile (that vary with frequency see full equivalent caps circuit)...a little step ,nice
Last edited:
I suggest to read this Nelson Pass paper about TDH profile 😉is all in audio 😎
https://www.firstwatt.com/pdf/art_h2_v1.pdf
https://www.firstwatt.com/pdf/art_Venerable_Triode.pdf
The last time I've done blind test with caps was one week ago. I have an active system and to protect the CD, I put a large electrolyte in series with it. Worst case scenario, can't hear a difference with or without.
However, I should obviously stick to gardening😀
However, I should obviously stick to gardening😀
Those particular results are contradicting, indicating something went wrong.
60% declared cap A sounds clear and cap B sounds dull, but other 30 % declared cap A sounds dull and cap B sounds clear! It can not be the both ways, it is impossible!
It is the same as to say that 60% of the people sees the colour of the grass is red, but other 30% sees the grass is green! Than, at least 30% (or whole 60%) of the people then suffer from daltonism, or are smoking too much "grass".
I don't understand you. Objectively and subjectively it is OK that 60% of the people heard the cap A is clear and cap B is dull, and that 10% can't hear any difference.
But, it is impossible that the other 30% heard the cap A is dull and cap B is clear! Someone is wrong - those 60% of the people, or other 30% of the people.
There could be any number of factors/reasons the preference/subjectivity differed, I thought that was why the greater requirement was for an objective, ie difference test. If we can focus on the audibility of a difference, for a moment, what would you consider proof?
They did hear a difference. If their opinion is wrong, does that really matter? All it means is not every one can be trusted to pick out the right one.
That shouldn't come as a big surprise if you look at this subjective test of full range drivers: https://www.diyaudio.com/forums/full-range/280725-subjective-blind-comparison-2in-3-5in-drivers-round-5-a-14.html#post4494341
The most votes went to the driver with the worst FR measurement. Now would you trust the masses to pick out stuff like this? That's what Toole and Olive use to determine favorites.
I will make my own choices, thank you. 😀
So, the worst capacitor has 0.00226% THD, and the best capacitor has 0.00064% THD. So what?!
Both distortions are way below human hearing threshold.
Perfectly logical explanation form post #1 is: 20% tolerance in capacitance is difference which can be easily heard.
Use any software for crossover design and you can easily see the 1-2 dB difference in tweeter output with 20% change in capacitance.
Toole's and Olive's work is NOT driven by science in the academic sense, it's a commercial pursuit funded by Harman/Samsung where Olive is employed and Toole is retired from.
They are not researchers or hold degrees in the field of cognitive hearing science, they are engineers and statisticians doing research funded by commercial interests, a fact both gentlemen repeatedly emphasize by the way.
Here are some actual researchers in the field that have not become minor celebrities promoting a brand:
Psychoacoustics and Cognitive Psychology (Past)
(PDF) Physiology, Psychoacoustics and Cognition in Normal and Impaired Hearing
Last edited:
Absolutely! The main target: to sell speakers. The main goal to achieve that? Please the masses.
While I value their research I won't forget the reasons why they do what they do.
It isn't used as my cookbook. I will make my own judgement.
While I value their research I won't forget the reasons why they do what they do.
It isn't used as my cookbook. I will make my own judgement.
It is much more complicated than the simple "They did hear a difference". Let me explain it with two experiments:
Experiment 1: Find whether cap A is different from cap B, not indicating why it is different.
Experiment 2: Find whether cap A is clearer or duller than cap B.
Results from Experiment 1: 90% of the people did hear a difference.
Results from Experiment 2: On the whole, 90% of the people did hear a difference. But, 60% of the people thinks cap A is clearer and cap B is duller, and other 30% thinks contrary - cap A is duller and cap B is clearer.
Although the same 90% of the people did hear a difference in both experiments, those two experiments are not identical. Can you spot the big problem with the Results from Experiment 2?
I have to quote myself (post #226):
It is the same as to say that 60% of the people sees the colour of the grass is red, but other 30% sees the grass is green! Then, at least 30% (or whole 60%) of the people suffer from daltonism, or are smoking too much "grass".
There is little argument that the threshold of audibility of distortion under some conditions is below 0.001% (-100dB) and approaching 0.0001% (-120dB).
As would be expected capacitor nonlinearities give rise to measurable THD in the frequency domain. But nonlinearities also affect time domain behaviour in a way to which THD measurements are blind, e.g. impulse response, step response and transient response aberrations.
Human auditory perception experiences audio stimuli in the time domain; in the frequency domain the ear is quite a blunt instrument. I suspect that time domain aberrations are what exposes differences in capacitor sound for different capacitor materials and constructions.
To ensure that jitter in digital clocks do not cause audible artefacts in digital audio reproduction it is necessary that timing errors are <0.00001% of the clock period, or <-140dB.
Sorry, it appears you are repeating yourself, but have not answered my questions in post 228
Inconsistent data is dealt with by statistical analysis, the purpose of which is to extract a valid signal from a noisey background. With a large enough sample size statistical analysis can determine whether the grass is green or red with a very high degree of certainty even though the observations recorded are inconsistent.
The data extracted by statistical analysis can be applied to the laws of physics and when it is, valid results are produced, something that random data does not do. To use your example, the statistics indicate that grass is most likely green. 60/30 is a large enough difference to not need to calculate standard deviations and confidence intervals to know the answer. Applying the physics of pigmentation it can be deduced that by adding red pigment to grass it will change its colour to yellow, where as after adding green pigment the grass will still be green. It won't surprise most people that red pigment would make grass yellow.
Of course some people will continue to see grass as red, or argue that grass is red, or claim that human eyes can't see the difference between red and green, or that grass can't change color, or that the color of grass doesn't matter, or whatever, all of which ignore the science. The equivalent of those arguments have played out in this thread LOL.
Last edited:
If a difference does exist and 90% can hear a difference, there is a difference.
If they can't see eye to eye which one is better, than that vote doesn't tell us anything. Just like picking out a cap to use could be a personal choice.
Now I would base that choice on measurements. If it measures like the simulated value, it gets my seal of approval.
Like I've shown in the full range test, the most voted driver was the one with the worst measurement. The lesson from that? Don't let an average crowd choose anything important. That's almost a political answer (lol).
If they can't see eye to eye which one is better, than that vote doesn't tell us anything. Just like picking out a cap to use could be a personal choice.
Now I would base that choice on measurements. If it measures like the simulated value, it gets my seal of approval.
Like I've shown in the full range test, the most voted driver was the one with the worst measurement. The lesson from that? Don't let an average crowd choose anything important. That's almost a political answer (lol).
Sorry, it appears you don't have arguments to contradict mine.
As for your previous question:
You have it fundamentaly wrong.
Difference heard from test subjects-people is not objective, but subjective test!
Proof for difference is ABX test (or any of it's variants), which is the test accepted as the ultimate proof, by science . And by me.
Preference test without stating what is different, is a weak one. Much stronger and better is a preference test with a range of predefined (or freely chosen) descriptions: clear vs dull, for example.
If the people can hear a difference, we are more than half way there...
There was a lot of doubt if that were even possible, what they prefer is an entirely different question.
There was a lot of doubt if that were even possible, what they prefer is an entirely different question.
- Status
- Not open for further replies.