Comparing audio gear is very difficult

[carewser]

Human memory is much shorter for things we hear than it is for things we touch or see but I never hear this mentioned in audio circles. I heard many years ago that accurate recall for sound is ~8 seconds which is why the only way to really compare gear is by doing A/B comparisons. I basically live in a little stereo shop because I own 8 speaker systems all hooked up and thanks to the magic of a couple of 3.5mm audio splitters leading from my laptop I can immediately A/B compare any of my 8 different speaker systems or have any/all playing simultaneously. When I get them all going, the volume could put a concert/nightclub to shame so it's kind of neat to have pretty much any volume imaginable on tap but when it comes to comparing sound though even when A/B comparing speakers, the thing I notice is that different systems not surprisingly, sound different from one another. The tough part is deciding which sounds better. I can't even imagine comparing the sound of different components. Again, I never hear anyone talking about this but it seems like an important point because other people seem to automatically and immediately know when they hear better sound but I rarely do. To me the only clear and defining difference between speakers has always been the depth of bass which I can clearly hear. Also, how do people intuitively know which sound is the way the music is supposed to sound? An ad for speakers a long time ago (I can't remember which ones) made a great point that unless you were in the recording studio with the musicians and sound engineers making the recording, there's no way to know exactly how any recording is supposed to sound and even if you were there, the sound recall thing comes into play again.

 

[Wes]

I never heard this mentioned in scientific circles. Where did you hear or see it?

 

[PierreV]

I never heard this mentioned in scientific circles. Where did you hear or see it?

Maybe you heard it more than 8 seconds ago? (sorry couldn't resist)

Abstract
Studies of the memory capabilities of nonhuman primates have consistently revealed a relative weakness for auditory compared to visual or tactile stimuli: extensive training is required to learn auditory memory tasks, and subjects are only capable of retaining acoustic information for a brief period of time. Whether a parallel deficit exists in human auditory memory remains an outstanding question. In the current study, a short-term memory paradigm was used to test human subjects’ retention of simple auditory, visual, and tactile stimuli that were carefully equated in terms of discriminability, stimulus exposure time, and temporal dynamics. Mean accuracy did not differ significantly among sensory modalities at very short retention intervals (1–4 s). However, at longer retention intervals (8–32 s), accuracy for auditory stimuli fell substantially below that observed for visual and tactile stimuli. In the interest of extending the ecological validity of these findings, a second experiment tested recognition memory for complex, naturalistic stimuli that would likely be encountered in everyday life. Subjects were able to identify all stimuli when retention was not required, however, recognition accuracy following a delay period was again inferior for auditory compared to visual and tactile stimuli. Thus, the outcomes of both experiments provide a human parallel to the pattern of results observed in nonhuman primates. The results are interpreted in light of neuropsychological data from nonhuman primates, which suggest a difference in the degree to which auditory, visual, and tactile memory are mediated by the perirhinal and entorhinal cortices.

 

[Jim Matthews]

Human memory is much shorter for things we hear than it is for things we touch or see but I never hear this mentioned in audio circles. I heard many years ago that accurate recall for sound is ~8 seconds which is why the only way to really compare gear is by doing A/B comparisons.

If this was true, we would not remember the name of singers, or the sound of instruments. I would hazard a SWAG that professional reviewers can recognize the sound "signature" of more familiar loudspeakers at least.

As the resolution of digital gear advances, I suspect it will become ever more difficult to identify differences between products.

As we cross that threshold, devices will impart less coloration and as you assert, become less memorable.

 

[Wes]

Maybe you heard it more than 8 seconds ago? (sorry couldn't resist)

Abstract
Studies of the memory capabilities of nonhuman primates have consistently revealed a relative weakness for auditory compared to visual or tactile stimuli: extensive training is required to learn auditory memory tasks, and subjects are only capable of retaining acoustic information for a brief period of time. Whether a parallel deficit exists in human auditory memory remains an outstanding question. In the current study, a short-term memory paradigm was used to test human subjects’ retention of simple auditory, visual, and tactile stimuli that were carefully equated in terms of discriminability, stimulus exposure time, and temporal dynamics. Mean accuracy did not differ significantly among sensory modalities at very short retention intervals (1–4 s). However, at longer retention intervals (8–32 s), accuracy for auditory stimuli fell substantially below that observed for visual and tactile stimuli. In the interest of extending the ecological validity of these findings, a second experiment tested recognition memory for complex, naturalistic stimuli that would likely be encountered in everyday life. Subjects were able to identify all stimuli when retention was not required, however, recognition accuracy following a delay period was again inferior for auditory compared to visual and tactile stimuli. Thus, the outcomes of both experiments provide a human parallel to the pattern of results observed in nonhuman primates. The results are interpreted in light of neuropsychological data from nonhuman primates, which suggest a difference in the degree to which auditory, visual, and tactile memory are mediated by the perirhinal and entorhinal cortices.

Sorry, posting something labelled "Abstract" is just not doing it for me.

now, once again, where did you see this?

and post the Methods section, so I'll know if I should bother reading the conclusions

 

[mhardy6647]

I googled the abstract text.

https://pubmed.ncbi.nlm.nih.gov/24587119/

looks like the full text and data are at:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3935966/

 

[carewser]

That Abstract mentioned that, ".....auditory, visual, and tactile memory are mediated by the perirhinal and entorhinal cortices"

Well duh

Not only did that prove my point about sonic memory being much shorter than other kinds but there's also this:
"Researchers at the University of Iowa have found that when it comes to memory, we don't remember things we hear nearly as well as things we see or touch." From: https://www.sciencedaily.com/releases/2014/02/140226174439.htm

Maybe you heard it more than 8 seconds ago? (sorry couldn't resist)

Good one

If this was true, we would not remember the name of singers, or the sound of instruments. I would hazard a SWAG that professional reviewers can recognize the sound "signature" of more familiar loudspeakers at least.

That's a good point although the memory required to know a singer's name is not the same as the memory required to know exactly what a pair of speakers sound like as sonic memory is much shorter. I expect we all know what drums, pianos and guitars sound like since they're baked into our minds as we have all heard them hundreds if not thousands of times

 

[Blumlein 88]

It is also called echoic memory. And no it doesn't mean we have no auditory memory at all after 8 seconds.
https://en.wikipedia.org/wiki/Echoic_memory

In essence the direct signals heard by our ears are stored and can be compared for a few seconds unadulterated. Then medium term working memory begins to process it for a few more seconds. Working memory does something like MP3 encoding does, and records a reduced resolution copy. Working memory processes and teases out various pertinent aspects of the sound that go into long term memory (which can last a lifetime). Long term memory however is not equivalent too nor can it be directly compared to echoic memory with much accuracy. Long term memory can also be effected by various other things beyond the direct sensual perception of the sound.

 

[carewser]

Wow, according to that wikipedia article echoic memory is only 3-4 seconds which is half as long as I thought so all the more reason to doubt people's perception of sound. I certainly doubt mine. All I really know is that when I hear deep bass in any music, I love the way it sounds and it brings a depth and fullness to music the way nothing else can

 

[PierreV]

Sorry, posting something labelled "Abstract" is just not doing it for me.
now, once again, where did you see this?
and post the Methods section, so I'll know if I should bother reading the conclusions

I am a bit confused by your reaction here. That echoic memory thing is the cornerstone behind the quick ABX testing usually recommended here and which you have seemed to frequently support (if I remember well), as opposed to the slow, long term listening comparing loudspeakers several months or years apart usually advocated by subjective audiophiles reviewers...

 

[Wombat]

If you are remembering your past comparisons, that is very different to comparing a current impression to those past ones.

 

[Jim Matthews]

That's a good point although the memory required to know a singer's name is not the same as the memory required to know exactly what a pair of speakers sound like as sonic memory is much shorter.

Some of us can recognize the voice over actor in commercials (John Cusack, for one - Fred Savage for another). I'm familiar with most of Ella's Fitzgerald's recordings and hear her clearly.

On hearing a large cat grumble, most of us will feel some measure of a fear response, even hair standing up on the back of our necks.

I think what you're grasping at is that our auditory sense is deeply adapted to the sound of speech, and threat recognition.

It hasn't been long enough for an adaptive change to recorded playback differences to be persistent in memory.

Therefore: what listening tests affirm (after spectrum sweep analysis) is how much the playback approximates naturally produced sounds.

We've been deviating from our original program since Bing Crosby stepped up to the first practical microphone.

 

[Wes]

I am a bit confused by your reaction here. That echoic memory thing is the cornerstone behind the quick ABX testing usually recommended here and which you have seemed to frequently support (if I remember well), as opposed to the slow, long term listening comparing loudspeakers several months or years apart usually advocated by subjective audiophiles reviewers...

That echoic memory thing is often misunderstood.

Without a link at minimum, or posting of Methods the abstract is completely uninterpretable.

 

[Blumlein 88]

https://www.healthline.com/health/echoic-memory#seeking-help

A more straightforward explanation for echoic memory.

https://www.magneticmemorymethod.com/echoic-memory/
A more involved explanation with active listening discussed and how active listening makes echoic memory available for us to use.

 

[txbdan]

Maybe its something like typical visual memory vs photographic memory. I can remember how a song goes, but I couldn't review the mix purely by memory.

 

[richard12511]

Some of us can recognize the voice over actor in commercials (John Cusack, for one - Fred Savage for another). I'm familiar with most of Ella's Fitzgerald's recordings and hear her clearly.

I think this is a bit different. The sound of Ella's voice is stored in your long term memory already, but it's stored at an extremely low resolution level. Luckily, human voices are unique enough to still be able to distinguish them at that low of a resolution. However, that long term memory resolution is somewhat inadequate for really comparing similar loudspeaker differences, and totally inadequate for comparing electronic component differences. If I let you listen to Ella's voice over a KEF and Revel speaker, A/Bed with instant switching, your short term echoic memory is sufficient to consistently identify which is which. However, if I let you hear Ella's voice over the KEF and Revel, then wait a year, then play you a sample, you'd have great difficulty telling me whether that sample is the KEF or Revel. Doing that with most electronic components would be impossible.

The idea that reviewers can hear a DAC or Amp they have atm and compare it against a DAC or amp they had years ago is very, very hard for me to believe.


The memory @carewser is talking about is not the super low resolution long term memory you speak of, but the high resolution memory that is required to compare electronic components.

Loudspeaker differences are easy to hear, and a somewhat lower resolution memory will do. However, I'm similar to @carewser in that while I can hear clear differences, I have a tough time deciding which of those differences I prefer.

I agree with you that we're likely especially good at remembering human voice differences.

 

[MOCKBA]

A good point. I guess a purpose of the site exactly do not ralay on human ears, but use a good quality equipment which can precisely tell you what is sounding better. Otherwise this site was useless.

 

[mhardy6647]

That echoic memory thing is often misunderstood.

Without a link at minimum, or posting of Methods the abstract is completely uninterpretable.

I posted the links to the article that google tied to that abstract earlier -- perhaps you have me on "ignore"?
https://www.audiosciencereview.com/...udio-gear-is-very-difficult.22250/post-737778

 

[Jim Matthews]

I think this is a bit different. The sound of Ella's voice is stored in your long term memory already, but it's stored at an extremely low resolution level. Luckily, human voices are unique enough to still be able to distinguish them at that low of a resolution.

(emphasis added)

I'm dubious of quantifying something without any basal reference.

Play twenty decent singers recordings, covering the same song.
Recognizing them by name seems a fairly demanding of memory..

Perhaps my stable memory is different, but thinking of pernicious Pop hits can bring most of the record to mind. This collapses with orchestral music, for me as there's so much going on.

 

[abdo123]

If this was true, we would not remember the name of singers, or the sound of instruments. I would hazard a SWAG that professional reviewers can recognize the sound "signature" of more familiar loudspeakers at least.

we remember melodies, which includes instruments and their harmonics.

but something like the reverberation of sound in a room? For example, if you're testing two sets of EQ for your room. Well, good luck lmao.