It is a rather unique application for an oil, the application calls for a low viscosity oil with low shear strength. I've been using Cramolin Blue, but your point is well taken, I'll have to research some...
Howie
There is one. The blind testing guys are right. The anti-ABX guys are also right.
Seems to me the problem is that the only testing listeners have tried is some form of ABX, which IME is insensitive, as Jakob correctly describes.
To come to some agreement on this issue, a new blind test computer program (for foobar, or otherwise) would have to be written and found to be acceptable by both sets of parties. My own experiments suggest that finding a mutually satisfactory solution should be possible, in theory at least. The big problem is getting a new program in our hands that we can try. If we had such a program, I think the perceptive listener group would find it useful for their own purposes. Here, I am thinking of using it for self-calibration of listening acuity.
If we don't have a good, reliable way to self-calibrate, we are likely to do what humans tend to do which is to become overly influenced by overconfidence bias and think we can hear more than we actually can. It happens. Just as overconfidence bias influence occurs on both sides of some of these arguments we have in this thread.
Last edited:
It was a fair generalisation which applies to almost all circumstances. People are easy to fool, especially those who think this does not apply to them.Max Headroom said:
I really don't understand how some believe bias can be ignored. Wait a moment - yes I do understand - it is just a slightly more foolish version of fooling yourself.mountainman bob said:
We did not invent the idea of unconscious bias. A priori we might expect that an amp which measures worse will sound worse; that fits in with our simple naive world view that better is better than worse. It was the psychologists and other psycho-xxx people who told us that people are easily fooled, so that worse can be better than better under some circumstances. I guess accepting the results of science is an agenda.Max Headroom said:
That is not the Sex Pistols, sorry to disappoint you. It’s called Public Image Limited.
Now back to your classic sales pitch 😀
Despite my intention not to post here anymore, and, with hindsight, stumble these sterile discussions about the sex of angels, I fall over-it. I burst out laughing so much it's true.
Dimitri, you made my day.
Moreover, wanting to reciclate old HITACHIs 2SK135 / 2SJ150 that I love, I spent the day yesterday to design a small 150W amp with them. It's a CFA with a diamond input, a cascode VAS, and a follower transmitter to drive a pair of these two power FETS.
Here my result for 150W (+-50V).
I used RPEREZ models I had that works ok, verified with Cordell models for 134/149 with close results.
I will post it on DIY when i will have real working amp with them, with real measurements and long term... Listening.
With a special dedication to my friend SYN08, whose elegance and intellectual honesty fills me with admiration. Quiescent current at 100mA ( Hitachi tempco ).
Oh, btw, if you ask about stability margin, i will give-you the Rolls Royce answer. Phase margin "sufficient" (>90°).
Attachments
Last edited:
Now that you unpack all that, I've never heard of an ABX or similar formal test involving long-term listening.
It's well established (and I've seen others state a few times in this thread, often in relation to ABX/DBT) that the best aural memory (I can hear arguments already, but "best" for at least some purposes) is short-term. In "The Loudspeaker Cookbook" the chapter on speaker testing describes two motorized turntables each with test speakers on them so a listener can quickly switch between speaker pair "A" and "B" to quickly listen to a short passage as reproduced by both pairs of speakers.
But I'm still not aware of any (scientific, reproducible) "hard observations" made from long-term listening. I'm certainly not saying such a thing doesn't exist, as I'm not that well read in audio research, but I've read various claims in this thread and I wonder if there's perhaps an AES article on the topic.
I can't argue with that. I recall when MP3 became popular (I wrote about this in this thread in recent months), when I first heard a 128k MP3 it sounded fine to me. I had to listen to a CD track encoded at different bitrates to get hints at what to listen for before I could hear the artifacts at 128k. One aspect of new formats is they often generate new types of distortion.
Well take me for example I have this (apparently rare)
ability to ignore or shut off everything except the goal set forth......if I know there's visual bias or better reviews or more cost involved I can compartmentalize these things so that theyre not a consideration in the final results.
Basically analyzing all the variables mentally without emotion.
It's not really that hard......in fact many times my results are surprising in the sense that I would have thought differently with just an educated guess.
This is a table excerpt from a sensory experiment (taste) which compares different test protocols:
(Yu-Ting Huang,Harry Lawless, Sensitivity of the ABX Discrimination Test, Journal of Sensory Studies 13 (1998) 229-239)
Yet another sample of Jakob(x) MO: conveniently jumping from particular to general conclusions.
I am sure everybody would appreciate details on the rationale of jumping from the data for "ABX can lead to false negatives in sucrose content in grape beverage" to the general conclusion (quickly extended to audio testing):
Why, thank you
I'm sorry,but you made a very strict assertion/statement combined with a question,it was:
To illustrate the issue I provided an example which explains, which way a difference can "automagically disappear under blind testing" .
Even when the sensory evaluation is a quite simple one (onedimensional and directional) while we are dealing here with multidimensional evaluations.
As you now know, how easily a difference can disappear (if it qualifies for "automagically" is open for debate 🙂 ) do you now accept, that it is not only belief, but something very real?
As I've written literally numerous times, the property "blinded" is a point (that must be included for being able to show validity) - internal validity to be more precise - but is otherwise just a minor point, as a plethora of other things have to be considered to ensure that a test is objective, valid and reliable .
Unfortunately in these forum discussions the "blind" propertiy is still presented as the most important one due to constant underestimation of the difficulties and due to simply neglecting/dismissing the available evidence.
Working on room acoustics is assessed by the degree of successful application. It's never complete. You can always improve. Your project plans for amps and DACs can have specs way beyond audible traits. What good will that do other than bragging rights?
In your opinion of course. I'm curious, what is your opinion based on?
Who said you aren't allowed to take your time in audio ABX?
It also performs a role in some very important ways in audio electronics too.
But but but..., that cripples listener's ability to pick out whatever they are hearing. They have to be allowed to peek so that they can trust their ears when evaluating audio electronics. 😀
So it was your stawman argument, as I suspected. Thanks for confirming.
How was your experience conducted? Oh, wait, it's classified.
Got it. 
TNT, as I have pointed out.
Does it matter?
I tried all the publicly available freeware ABX applications. Foobar ABX is the most developed. None of them are any more sensitive than it. At least foobar ABX allows one to type in starting and ending times of long files so that short segments can be compared. Also, foobar ABX plays both sample files at the same volume level. Whether or not they come out as level matched depends on the files themselves.
If we are going to measure levels for matched comparisons, in most cases I would probably favor using a perceptual loudness VU meter over RMS, peak, average, or other level measurement methods. The best loudness metering plugins money can buy - and my favourite alternative (for only 9 dollars !) - Production Advice
https://www.itu.int/dms_pubrec/itu-r/rec/bs/R-REC-BS.1770-4-201510-I!!PDF-E.pdf
- Status
- Not open for further replies.