That makes sense actually, I can visualise what happens if a starting point is actually at a slightly different level.
An idea then... something like a short 5 or 10 second music sample of something like a brushed cymbal, something with lots of detail and shimmer, and with a quick fade in and out of say 0.5 to 1 second. I imagine that would remove the need for super precision in alignment.
So maybe we can say that if you cannot be sure of a difference with a 0.5 sec delay but with exact amplitude levels, then it isn't really worth worrying much about any differences !
That would mean we can go ahead with totally enjoying the music ( and the
) and not be worried about missing any cymbal brushes etc ! 🙂
That would mean we can go ahead with totally enjoying the music ( and the
) and not be worried about missing any cymbal brushes etc ! 🙂I´m still surprised by the categorical conclusions that aren´t really justified by these results (due to the test design). 😉
Maybe i missed the complete description of the setup and the research hypothesis.....
Would you try this one, then? I think you might be satisfied. It is short, dynamic. You would not find it in a pro release.
http://pmacura.cz/drumhats.zip
DR = 19 🙂
Attachments
Last edited:
The delay example is imo more related to differences that can be perceived in a test although being totally irrelevant to the sound quality.
Overall your conclusion seems to be reasonable but unfortunately we already know since a long time ago from several published experiments that even quite big differences can remain undetected in tests...
That would mean we can go ahead with totally enjoying the music ( and the
Combining a system that allows to enjoy the music should be the first priority, looking afterwards for thing that might be even more enjoyable can be part of the hobby too. 🙂
Easy to overcome. (just like preventing type 2 errors, unlike what others in this thread have claimed.)
If someone has proven to hear a difference between some files, for instance get 19 out of 20 correct in an ABX test, then he/she can learn others what has been the audible clue. Others can then focus on this clue and see if they can also get a positive outcome of the test.
This method (its just training of listeners) is used all the time in more scientific studies.
Tip: to get sample accurate files you must use wordclock. Most pro level soundcards have this option, don't know any CD players that have this though.
Or even better/easier use a soundcard that can play and record at the same time, most soundcards have this option.
This was used in the previous test. But not in this one, because the soundcard DAC would reduce quality of the 96/24 files used in this test. It is not to be debated, because I have all my components thoroughly measured.
So we have now 3 - 8 us differences between the sampling of the files. Probably nothing too bad. So far, no one has posted a valid ABX result that would prove he heard the difference. Just unsupported impressions 🙂
A quick listen and that sounds good. I would say that would make a good test sample.
Thank you 🙂. It was recorded by a friend of mine. He has excellent recording equipment and I love his raw, unmodified first takes without any manipulation with amplitude, dynamics etc.
The question is if anyone is still interested in such test. I have now limited time till Saturday and then I will be 1 week off.
I was going to say your playing had improved Pavel 😉
Now I know how much effort goes into preparing these tests and so I would say that it would be great if you could... but lets see what interest there is for another test.
Would you envisage using the same test buffer configuration ?
Now I know how much effort goes into preparing these tests and so I would say that it would be great if you could... but lets see what interest there is for another test.
Would you envisage using the same test buffer configuration ?
It would be the easiest way for me, and 3rd test in row to confirm or disprove previous results. As a "previous result" I mean no positive ABX yet 🙂
That sounds like a plan. Consistency in method and all that. If you feel you can devote the time to it then I'm sure it would be appreciated.
Pavel,
My opinion is that at higher gains one must be more selective of component choices. With CDs or virtually any digital storage media one can decode the data at a high enough level that the only gain required would be in the audio power amplifier.
What is unresolved is the gain stages required to record the acoustical source. To get the cat chasing it's tail how can one determine if there is a flaw in the recording process that is more acceptable if masked in the playback tests?
(Hint use multiple sources.)
Many years back the most popular studio monitor loudspeakers showed nice flat on axis frequency response when measured with a precision 1" microphone capsule. Using a more modern smaller capsule showed they had a high frequency roll-off.
As home reproduction loudspeakers were often final voiced by ear, music mastered on the newer better studio monitors was often derided as being too shrill.
My opinion is that at higher gains one must be more selective of component choices. With CDs or virtually any digital storage media one can decode the data at a high enough level that the only gain required would be in the audio power amplifier.
What is unresolved is the gain stages required to record the acoustical source. To get the cat chasing it's tail how can one determine if there is a flaw in the recording process that is more acceptable if masked in the playback tests?
(Hint use multiple sources.)
Many years back the most popular studio monitor loudspeakers showed nice flat on axis frequency response when measured with a precision 1" microphone capsule. Using a more modern smaller capsule showed they had a high frequency roll-off.
As home reproduction loudspeakers were often final voiced by ear, music mastered on the newer better studio monitors was often derided as being too shrill.
Ed, of course, but there are many circuits with opamps connected as unity gain buffers (impedance converters) or with minimum gain like 2-5x. As you know, the 1458 (741), or AD712 (little better 072) were demonised many times here. You do not want for me to remind the thread. I am not saying they are SOTA parts, especially 1458, but, on the other hand, one would expect their trace will be clearly recognized, right? We now have even the source file here (1st post of the thread), but not a single positive ABX result. Something's wrong? 😛
For higher gains or microphone preamps, no one sane would use the 1458. So it is not a goal to give it 100x gain and 50ohm load. Engineering goal is to use the part inside the range of its specs, without performance degradation.
For higher gains or microphone preamps, no one sane would use the 1458. So it is not a goal to give it 100x gain and 50ohm load. Engineering goal is to use the part inside the range of its specs, without performance degradation.
Probably you shouldn't be surprised, these are not scientists. They are mostly hobbyists with a lot of preexisting opinions that don't necessarily change from one listening test. It takes lots of time and tests to gradually sway most people's beliefs, if they budge at all.
Last edited:
Independently replicated experimental studies are probably the gold standard for discovering scientific truth, at least in medicine and social sciences.
But, even at that, when preexisting beliefs are strong it can be very difficult and take a long time to for many people to become convinced they have to abandon their old, sometimes strongly held beliefs.
On the other hand, (1) when people don't have preexisting beliefs, (2) they believe a well informed audience will judge their findings, and (3) they know the audience wants the truth, then most people will try hard to be objective .
Last edited:
- Status
- Not open for further replies.