[Darren G]

As you should. After all, you are human.

Your brain ignores vast amount of data your ears capture. Can you imagine storing every sound around you with every nuance in each? Can you remember every note in every piece of music you hear?
...

^^^ This. Funny thing is nobody is surprised that our brains throw out vast amounts of other information (visuals, memories, etc.), for all the same good reasons

--

Tangentially related to this point, but it's worth searching for and taking a listen to the MP3 samples of Tom's Diner. If I can find the web page again I'll link it, but what is interesting is to listen to A.) The original uncompressed sample, B.) The MP3 compressed sample, and C.) The difference between the A and B samples. Thing is, there clearly are differences if you listen C. Still, even listening to that difference sample, and immediately switching back and forth, I sure cannot hear those differences when switching between A and B. Evidence that it's not just that our brain's forget details, but that our brains are actively filtering auditory information all the time.

Then again to me I just expect that my hearing, and visual processing are not 'perfect' (i.e., flawless measuring tools), but they are both good enough to capture what is of value, while tossing what is useless

 

[gregorio]

Amir is merely stating what has been demonstrated about the accuracy of the stored memory.

Maybe I misunderstood what he was saying, maybe he meant that what we analyse is entirely captured in echoic memory but his statement that echoic memory is "very high fidelity" and "effectively captures everything" I took to mean that everything is captured, not just what I happen to be concentrating on/analysing. If it's the latter, then I want to learn more about it. If it's the former, then it's just my misinterpretation and there's nothing new here (for me personally).

In case I didn't make it clear: I'm not at all disagreeing with his post, on the contrary I officially "liked" it. I haven't seen that poster before, he/she seems to be a new member and in my experience here, such an enlightened post, a post detailing accurate (and APPLICABLE!) facts and apparently unburdened by audiophile marketing myth is quite a rarity from a new poster! I'm NOT trying to pick holes in what he/she stated, despite the fact it may appear that way. Although my work requires considerable practical experience in understanding and taking advantage of the quirks of human hearing/perception, I certainly have some holes in my knowledge of the academic research and theories in this area, particularly the more recent ones.

G

 

[castleofargh]

I see, maybe I misread things too ^_^.
Amir is indeed new only as this account here. he has his own internet sandbox with audio and science in the title .

 

[amirm]

^^^ This. Funny thing is nobody is surprised that our brains throw out vast amounts of other information (visuals, memories, etc.), for all the same good reasons

--

Tangentially related to this point, but it's worth searching for and taking a listen to the MP3 samples of Tom's Diner. If I can find the web page again I'll link it, but what is interesting is to listen to A.) The original uncompressed sample, B.) The MP3 compressed sample, and C.) The difference between the A and B samples. Thing is, there clearly are differences if you listen C. Still, even listening to that difference sample, and immediately switching back and forth, I sure cannot hear those differences when switching between A and B. Evidence that it's not just that our brain's forget details, but that our brains are actively filtering auditory information all the time.

It is indeed. Lossy encoders work on psychoacoustics basis and assuming there is enough bit budget, will only throw out what is considered to be inaudible. Masking effect accounts for a lot of that where any tone casts a shadow around it where we can't hear other tones in that band at lower amplitude. Take out the main tone and then we do hear them as in (C) example you mention. That is why differential/null testing is unfair and improper in the case of lossy codecs.

Note that with training, some of the default psychoacoustic effects can be undone. I have such training for lossy codecs and hear things like "pre-echo" far better than general population including audiophiles. Likewise people who mix and record music and musicians tend to hear the effects of room reflections far more than general public.

 

[amirm]

So "lossy encoding" is a biological standard. Hmmm. Where can I sell this new information?

Unfortunately it is already in a dozen WIkipedia pages so no money left to be made at this point.

 

[amirm]

Now that's an assertion I find interesting, as it appears to contradict my experience. I personally find that I cannot capture everything, even for just a few seconds. If I want to analyse something in real detail I need several passes, even for very short passages, each pass focusing my concentration in different areas because I'm unable to consciously analyse/process all of it in one go. I was originally a formally trained and then a pro orchestral musician and then, for the last 25 years, a pro sound and music engineer/producer. I mention this only because my many decades of almost continuous formal and informal listening training probably results in me processing/analysing what I'm critically listening to somewhat differently to those without that experience.

It maybe that I've misinterpreted your assertion and what you were trying to say entirely agrees with my experience but if it doesn't, do you have any reliable sources you could pass my way, preferably a scientific paper or a white paper from a respected industry org? I'd like to make it ABSOLUTELY clear that I'm not looking for ammunition to start an argument, I genuinely want to learn more, out of both pure curiosity and because this line of enquiry could potentially lead me to a more efficient and/or effective method of critical listening analysis which would benefit me professionally.

G

The notion of dual stage memory concept (at high level) is verified by countless studies. See the references at the end of this Wiki page for example: https://en.wikipedia.org/wiki/Short-term_memory

The precise amount of that we store in the echoic memory is not easy to determine and it seems to depend on the domain/test being conducted. A lot of tests for example are for people to remember words, letters, etc.

My experience in audio comparison mirrors yours actually depending on domain. For example in double blind tests of speakers at Harman with switchover time of 4 seconds and hearing a clip for 30 seconds or so, I could tell the fidelity of the speakers apart with ease. In the test of 320 kbps MP3 against the original, the switchover time better be in sub-seconds or I can't find or hear those differences.

It seems that the whole notion of switching between two stimulus is the enemy of short-term memory. Any glitches, time lapse, etc. can cause the memory of the old to be erased or faded.

Also, when we do such testing, we are bombarded with acoustic data, much of it may be the same in A and B samples. Work then is involved in isolating what is different and remembering that as to make a decision.

So in detail, there is a lot more complexity. The original point remains though: that we influence what we choose to remember about what we hear. We have to as we can't capture it all and remember for longer term. This act causes us to hear things that before were forgotten from long-term memory. It is those finer details that we associate with high-fidelity making us routinely think the new gear sounds better where in reality all we did was choose what to remember.

 

[bigshot]

For sound, the more similar the sounds, the shorter the memory. And of course, when we're comparing high fidelity audio equipment and codecs, we're talking about very very similar sounds.

 

[gregorio]

[1] The notion of dual stage memory concept (at high level) is verified by countless studies.
[2] The precise amount of that we store in the echoic memory is not easy to determine and it seems to depend on the domain/test being conducted.
[3] For example in double blind tests of speakers at Harman with switchover time of 4 seconds and hearing a clip for 30 seconds or so, I could tell the fidelity of the speakers apart with ease. In the test of 320 kbps MP3 against the original, the switchover time better be in sub-seconds or I can't find or hear those differences.
[4] Also, when we do such testing, we are bombarded with acoustic data, much of it may be the same in A and B samples.
[5] The original point remains though: that we influence what we choose to remember about what we hear. We have to as we can't capture it all and remember for longer term. This act causes us to hear things that before were forgotten from long-term memory. It is those finer details that we associate with high-fidelity making us routinely think the new gear sounds better where in reality all we did was choose what to remember.

1. Yep, I wasn't questioning the notion of dual stage memory.
2. OK, it seems I misinterpreted what you meant. I thought you might be aware of some reliable evidence of which I was unaware.
3. Again, yes, I agree. It seems to me that the smaller the magnitude of the difference/s and the more concentration and tight focus is required to discern it, the less likely it is to be able to accurately remember that difference for more than a very few seconds. A situation I experience several or many times a day, every day!
4. Are you talking about acoustic data in the listening environment or actually recorded in the samples? Either way, I agree, just curious. It's an area largely ignored by many audiophiles, whereas for me it's an area in which I frequently have to work, actually having to create that acoustic data and an area where I've spent considerable time and money trying to reduce (in the case of my listening/monitoring environment).
5. I wasn't questioning your original point at all, in fact the opposite. It's a point which is rarely considered, if at all, and is part of a group of important points which many audiophiles really need to be aware of, IMHO.

G

 

[castleofargh]

speaking of focus and consciousness and experience of what we're listening to, the other day I was trying to find a famous situation to demonstrate some of it. we listen to some garbage highly degraded signal, next we get to hear a sentence(the non degraded signal) once. and then we can't help but recognize that sentence when we listen to the degraded signal again. somebody knows the name of this effect/principle? my google-fu failed me.

 

[bigshot]

That's "power of suggestion". It was a big part of the backwards masking craze

 

[Redcarmoose]

Now that's an assertion I find interesting, as it appears to contradict my experience. I personally find that I cannot capture everything, even for just a few seconds. If I want to analyse something in real detail I need several passes, even for very short passages, each pass focusing my concentration in different areas because I'm unable to consciously analyse/process all of it in one go. I was originally a formally trained and then a pro orchestral musician and then, for the last 25 years, a pro sound and music engineer/producer. I mention this only because my many decades of almost continuous formal and informal listening training probably results in me processing/analysing what I'm critically listening to somewhat differently to those without that experience.

It maybe that I've misinterpreted your assertion and what you were trying to say entirely agrees with my experience but if it doesn't, do you have any reliable sources you could pass my way, preferably a scientific paper or a white paper from a respected industry org? I'd like to make it ABSOLUTELY clear that I'm not looking for ammunition to start an argument, I genuinely want to learn more, out of both pure curiosity and because this line of enquiry could potentially lead me to a more efficient and/or effective method of critical listening analysis which would benefit me professionally.

G

That has always been something that I have found interesting when reading here. The fact may be that the brain can not hold the true character of the quality of the reproduction for very long in memory, hence disqualifying the relevance and accuracy in side by side listening tests.

The simple fact that subjective side by side listening is both a practical yet a flawed and futile way to judge fidelity. And as mentioned, the fact that the brain may be concentrating on only one aspect of the music and not be able to fully ascertain the whole signal, that it takes multiple listening (of a musical passage) to focus in on the details. As in many cases were not just concentrating on one thing in the music but and entire cornucopia of traits and qualities; sound-stage, treble, mid-range, bass quality and detail, timing, pace, placement of imaging and on and on. Just the fact that some equipment even could in fact contain momentary-responce-hiccups where it does everything right, but then drops the ball in one response category for just a moment.

Yet I tend to also feel it's the overall effects of a long listening experience and the overall intrinsic-long-term entertainment quota which could in fact be the information to determine the quality of the equipment in question. It's like a dinning experience where the sum of the whole out-weights the small details. Simply listening to a 5 second sound-bite is not going to replicate the emotional experience contained upon listening to an entire piece of music.

 

[gregorio]

[1] The simple fact that subjective side by side listening is both a practical yet a flawed and futile way to judge fidelity.
[2] Yet I tend to also feel it's the overall effects of a long listening experience and the overall intrinsic-long-term entertainment quota which could in fact be the information to determine the quality of the equipment in question.
[3] Simply listening to a 5 second sound-bite is not going to replicate the emotional experience contained upon listening to an entire piece of music.

1. Yep, which is why we don't use listening tests to judge fidelity!
2. So you're saying that having less information available, as is the case with long term memory, is "the information to determine the quality of the equipment"?
3. The "emotional experience" is your brain's response to the sensory data it's receiving, it only exists in your brain, it's not an intrinsic property of sound itself and it's not recorded or reproduced. Therefore, "emotional experience" is irrelevant when comparing recording or reproduction equipment!

G

 

[amirm]

On efficacy of short term vs long term testing, that has been formally tested. See my AES paper digest here: https://audiosciencereview.com/foru...ity-and-reliability-of-abx-blind-testing.186/

Summarizing, audiophiles who used long-term testing demonstrated far less ability to hear distortions that were provably where there, than those who did in an ABX box using quick switching.

I can attest to this personally. I have passed countless difficult tests of audibility by performing quick switching and sampling in some cases less than one second of music. Give me the same test in "long term" and I would be lost.

 

[Redcarmoose]

1. Yep, which is why we don't use listening tests to judge fidelity!
2. So you're saying that having less information available, as is the case with long term memory, is "the information to determine the quality of the equipment"?
3. The "emotional experience" is your brain's response to the sensory data it's receiving, it only exists in your brain, it's not an intrinsic property of sound itself and it's not recorded or reproduced. Therefore, "emotional experience" is irrelevant when comparing recording or reproduction equipment!

G

On efficacy of short term vs long term testing, that has been formally tested. See my AES paper digest here: https://audiosciencereview.com/foru...ity-and-reliability-of-abx-blind-testing.186/

Summarizing, audiophiles who used long-term testing demonstrated far less ability to hear distortions that were provably where there, than those who did in an ABX box using quick switching.

I can attest to this personally. I have passed countless difficult tests of audibility by performing quick switching and sampling in some cases less than one second of music. Give me the same test in "long term" and I would be lost.

Listening tests are untrustworthy. The factors I feel have to do with the subjective interpretations each different listener perceives. Add to that how each person remembers what they have just perceived. Tests on visual cognitive (episodic visual memory) show that the brain files images away in accordance to previously understood visual parameters. Thus that tea pot you saw in the last room may be remembered as having a round knob on the lid simply due to the fact that most tea pots have round knobs, when reality proves the knob in truth was square.

But from the earliest of audiophile days, quality has been based on how true to life the reproduction is. And of course there are scientific tests which show us when an audio response differs from what is regarded as the correct baselines. This is how we got to where we are!

Could distortions in the right places make a headphone, amp or DAC sound more musical? Yes.
Could there be finite responses from gear that still remain slightly out of reach of being quantified by measuring equipment?
And finally I would rather live with a technically imperfect system which allowed me hours of casual listening over a perfect specification reproduction that left the music lifeless and cold.


But in ending the proof comes with time. Obviously the colored system is going to get boring over time once the magic wears off and the brain figures out that there are frequencies enhanced or left out in listening. But I tend to look at it like cars. Car engineers use the cutting edge of computer design combined with the complete science and physics history of mankind. They have big bucks on the line and all the money in the world for R&D. Yet somehow we have models and year makes of great cars by different manufactures and cars that come out either normal or inferior. Cars now in relationship to cars from the 1960s are completely different machines. The level of tolerances and intelligence of design have made today's cars faster, more comfortable and finally more dependable than car makers of the 1960s could ever have dreamed of. The designs tested right before anything went into production.

Still it's the daily use of that car........the humanity of that new car.......it's personality in the end so to speak, which over time creates a relationship. It's this relationship which provides the consumer with how good the product is. Electrical equipment for the reproduction of audio follow the same parameters in the end.

 

[bigshot]

Listening tests are untrustworthy.

You don't trust your own ears?! Do you trust your philosophy and biases more than your ability to hear?