Not at all. Anybody can easily start to learn to improve listening skills. There are a range of abilities, it's not a binary thing. It's like anything else people do, there is some combination of natural ability and of learning and practice.
Hi, thanks for a sane postI wonder, as someone in the "business", do you consider yourself to be more a musician than an audiophile?
Hi Scott,
Not to dodge your question, however, I think it'd great if more musicians were audiophiles & conversely if more audiophiles were musicians
.I've worked/played on so many systems and continue to be engaged in most elements of "the business". It's vital to walk the line if you want the perspective. Lately my focus is exploiting new speaker topologies with the prolific XRK971 (resulting in an issued patent even.)
FWIW, the musical projects that interest me are hard to capture and don't have traditional spectral balance or mixing techniques and are more about concept & aesthetics. I'm drawn to mixes that don't conform or are even hard to playback correctly. The long standing trend seems to be everything (art included) is further organized into linear networks of responsibility and expectation for the purpose of 1>efficiency and 2>providing brokers a hook through which they can mediate experience for profit. None of that activity has a creative point. Play isn't about indigenous instruments only being used for the strings that are resonant to the western 12 tone scale so that artist can chart on spotify or get booked. Conformity sucks the energy out of systems. Literally. And sure, often it sounds better.
.
Last edited:
Before we go too far in talking about drawing conclusions from testing, maybe we could talk about to set up and conduct a test to deal with some issues.
Because of how the brain processes listening, some things probably ought to be talked about in terms of software to use.
Actually, Foobar ABX is almost good enough. It allows a listener to type in the start time and end time of the segment they want to listen to, although it is poor for browsing to find a good segment. Audacity or similar could be used to help with that, find start and end time of segment with Audacity then type them into foobar.
Problems with Foobar are that (1) it should have a checkbox to loop. It also requires looking to switch which sample is playing. So, (2)There should be one button to toggle between samples so it can be operated unsighted while listening.
If those two things were added to Foobar it might actually be quite usable for pretty low level testing.
Because of how the brain processes listening, some things probably ought to be talked about in terms of software to use.
Actually, Foobar ABX is almost good enough. It allows a listener to type in the start time and end time of the segment they want to listen to, although it is poor for browsing to find a good segment. Audacity or similar could be used to help with that, find start and end time of segment with Audacity then type them into foobar.
Problems with Foobar are that (1) it should have a checkbox to loop. It also requires looking to switch which sample is playing. So, (2)There should be one button to toggle between samples so it can be operated unsighted while listening.
If those two things were added to Foobar it might actually be quite usable for pretty low level testing.
The next thing that is a known issue is that the listener needs some training and practice with the testing. What often happens is that in normal listening System 1 processes in the brain are doing most or all of the heavy lifting automatically and effortlessly.
When someone is tested without proper preparation what happens is they automatically recruit System 2 processes to try to consciously analyze what they are listening to and for. People who know a little about frequencies start listening for frequency differences between samples. Essentially, they stop letting powerful System 1 processes do their work and substitute far weaker System 2 processes. Training and practice among other things serve to allow the listener to let go, relax, and just listen without thinking about it consciously at all. (Which is probably what they were doing if and when they said they could hear some difference earlier prior to testing).
When someone is tested without proper preparation what happens is they automatically recruit System 2 processes to try to consciously analyze what they are listening to and for. People who know a little about frequencies start listening for frequency differences between samples. Essentially, they stop letting powerful System 1 processes do their work and substitute far weaker System 2 processes. Training and practice among other things serve to allow the listener to let go, relax, and just listen without thinking about it consciously at all. (Which is probably what they were doing if and when they said they could hear some difference earlier prior to testing).
There is nothing unique about ABX testing of audio components (a form of statistical analysis) which would preclude extrapolation.
If there were, your reply would have likely been more than "we cannot."
But there is. It is in the statistics. The scientific facts of the test If the null hypothesis is rejected it only applies to that particular test/day/session.
dave
Yes.
There is skill, learned skill in consciously switching between System 1 and System 2 listening modes.
Try out your listening mode skills - Test Files
Dan.
Hi Dave,
The problem here doesn't seem to be exactly where you are focusing. They don't want to use ABX to find out if someone can hear something. They want to use it to prove someone couldn't hear something if they tried. They would like the null hypothesis to be that the test subject *can* hear something. And, they believe that they will have disproved the null hypothesis by having the test subject fail to hear something.
Of course, there is still a problem, but a different one as I assume you are aware.
EDIT: For those interested in what the null hypothesis is and why it's used:
Null hypothesis - Wikipedia
Its use is taught in statistics and elsewhere, but exactly why is an interesting question. Why not just try to prove what one wants to prove rather than trying to more or less disprove the opposite? One view is expressed here:
The Founder’s Null Hypothesis
I tend to agree. Its a good idea and the best way to go so as to help minimize the chances of fooling one's self.
Last edited:
Assuming we are all humans then we are all biased.
Regarding bias in this thread, there has been bias to the effect people can't hear all that much, and bias that some people can hear more than you might think.
On the "can't hear all that much" side, some seem to have a wish to prove their suspicions correct. That's perfectly fine, of course. Just run tests in a way designed to get reasonably accurate results for whatever listening skill is being test for.
To do that, it's necessary to understand about Naive Realism, the idea that our senses provide accurate depictions of the world around us. In reality, our senses are highly filtered and processed by System 1. We think we hear what there is and nothing else exists, but we don't. That has been proven by lots of research. From the perspective of Naive Realism it's easy to take the next step: Since we feel our senses are accurate as they are, there can be no benefit to training and practice.
It's also necessary to understand certain things known about aural perception and to be aware of some of what is still hazy or unknown. On the other hand, assuming one needs no more than common sense would be expected to easily lead to errors. That's particularly true when statistics is needed to interpret results.
Not exactly. They are to help prevent alternate explanations for test results, but some experiments don't employ them.
Positive controls are considered a part of good experimental design:
Scientific control - Wikipedia
For my own use, I use the Reaper DAW. I use it to find a short segment of music to loop, and set it up to compare two tracks with one-click switching. Also use it for volume level matching.
Then I either look the other way or close my eyes when I don't want to know what I am listening to. I click the mouse really fast a bunch of times while focusing attention elsewhere so I don't know which track is playing. Then I let one loop several times then switch to the other track. Usually it is obvious right away the other track is different, but not always.
Sometimes it takes some very careful listening to one track and the other to really latch onto the differences. Then the task is the make that lock-in process work reliably enough for bullet-proof testing. One error is one too many when testing, so practice until ready for test.
Then I listen blind and make a prediction about what I will see on the screen when I look. If I have chosen the correct track, I repeat a few times to be sure it wasn't chance and I am certain I can hear the difference reliably.
If I get it wrong, I go back and keep practicing until I can do it right every time.
One problem is that in some cases there is a huge amount of information I am listening to that is going by fast, and auditory memory fills up quickly. In PMA's Toccatta violin files test, it was only very subtle difference in the sound of individual inflections of finger vibrato over a few seconds that I picked to listen to. In one file the volume level of the vibrato inflections was more uniform, and there was a tiny bit more clarity that probably wasn't really clarity but that's what it sounded like in terms of the mental experience. It may well have been a little bit of harmonic distortion, as 3rd harmonic in very low doses can make things sound more clear in a good way. It took me a few tries to lay my finger on what was different about them, and a bit more practice to get it right every time. However, i would say that was a very difficult test. Probably about the hardest I can still do at my age.
If there were two people available, it would be better to let one run the computer out of sight of the other person who is listening once it's time to test. For practicing and for finding a segment to loop, the listener should be able to run the computer if they like.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.