Researcher jumps into hi-res discussion and owns hater.

[new_media]

I was very entertained by this high-res discussion on the Sonos forums:

 

https://en.community.sonos.com/ask-a-question-228987/can-digital-audio-playback-be-improved-by-resolutions-greater-than-16441-6765865/index3.html

 

A meta-analysis which showed small but perceptible differences between hi-res and redbook was mentioned, and a quite belligerent non-believer accused the author of the study of bias, lying, conflict of interest, etc. He then goes on to say that any differences are caused by intermodulation distortion, and hi-res therefore actually sounds worse.

 

We'll, the author of the study jumps in to set him straight on all points, answers a couple of his questions, and then flat ignores his apoplectic responses. Perfectly handled.

[Ralf11]

hmmm....

 

"My understanding of human memory of sound quality is that it degrades in seconds and becomes unreliable. To overcome that, a rapid - less than a second - changeover between the tested samples is required to get a valid comparison outcome."

[Ralf11]

Ha! one poster also confuses "scholarship" with scientific research

 

- good way to start a riot in a faculty meeting

[STC]

hmmm....

 

"My understanding of human memory of sound quality is that it degrades in seconds and becomes unreliable. To overcome that, a rapid - less than a second - changeover between the tested samples is required to get a valid comparison outcome."

 

Sometimes, when changes occurs seamlessly, their difference are not readily perceivable.

[Jud]

hmmm....

 

"My understanding of human memory of sound quality is that it degrades in seconds and becomes unreliable. To overcome that, a rapid - less than a second - changeover between the tested samples is required to get a valid comparison outcome."

 

Echoic memory has been found to range from about 4-10 seconds in most research, usually tending toward the shorter end of the range. There are also confounders that can result in failure to recall sounds even within that brief time frame. But "less than a second" is probably a little short.

 

 

Sent from my iPhone using Computer Audiophile

[STC]

I need help. How is echoic memory and time of changeover relevant?

[Jud]

I need help. How is echoic memory and time of changeover relevant?

 

If you take longer than 4 seconds or so to switch, you will not have an accurate memory of many of the characteristics of the previous sound to compare to the new one. Loudness does appear to be retained in memory for a longer period.

 

 

Sent from my iPhone using Computer Audiophile

[Miska]

If you take longer than 4 seconds or so to switch, you will not have an accurate memory of many of the characteristics of the previous sound to compare to the new one. Loudness does appear to be retained in memory for a longer period.

 

I recognize voice of family members longer than 4 seconds after they've shut up.

 

My DACs are similar family members who's voice I recognize. Same goes for my digital filters.

[esldude]

I need help. How is echoic memory and time of changeover relevant?

 

to add to Jud's response. Your brain seems to keep some data on hand that is the actual perceptual signal for a few seconds. Then it shuffles that off to short term and then long term memory. These other steps don't keep all the data or all the signal. So those memories of the sound are in some ways like an MP3 which discarded some of the genuine data for compression purposes. So if you tarry too long between short segments you don't really have the high quality actual perception to compare to the current actual perception. You are comparing a quality degraded (and perhaps quality modulated) memory to the real perception.

[Jud]

I recognize voice of family members longer than 4 seconds after they've shut up.

 

My DACs are similar family members who's voice I recognize. Same goes for my digital filters.

 

Yes, I absolutely agree. You are not using echoic memory for any of those things. Nor are you using echoic memory when, for example, you recognize the song John Coltrane is playing on a sax rather than singing, in different tempi and keys, is "My Favorite Things."

 

 

Sent from my iPhone using Computer Audiophile

[elcorso]

+1

 

Roch

 

I recognize voice of family members longer than 4 seconds after they've shut up.

 

My DACs are similar family members who's voice I recognize. Same goes for my digital filters.

[Ralf11]

to add to Jud's response. Your brain seems to keep some data on hand that is the actual perceptual signal for a few seconds. Then it shuffles that off to short term and then long term memory. These other steps don't keep all the data or all the signal. So those memories of the sound are in some ways like an MP3 which discarded some of the genuine data for compression purposes. So if tarry too long between short segments you don't really have the high quality actual perception to compare to the current actual perception. You are comparing a quality degraded (and perhaps quality modulated) memory to the real perception.

 

or maybe the feature extraction your brain does is adequate to resolve differences in the sound of A vs. B...

[STC]

Thanx guys! I just wanted to be sure I understood the earlier post because I wasn't sure of Ralf's quote.

 

Going back to echoic memory, I think there are instances that without intervals between samples you would not able to to tell the difference.

[Jud]

or maybe the feature extraction your brain does is adequate to resolve differences in the sound of A vs. B...

 

The research results are quite consistent. If there was "feature extraction" creating exceptions to failure of recollection of many sonic characteristics after 4 seconds or so, the tests should have different results.

 

But folks should really pay attention to what Miska said....

 

 

Sent from my iPhone using Computer Audiophile

[esldude]

or maybe the feature extraction your brain does is adequate to resolve differences in the sound of A vs. B...

 

Research into the matter doesn't mean no differences can be discerned longer term. Simply that the fineness of perceptions or in perceiving smallest differences degrades substantially. Not unique to hearing either. You have similar differences related to all your senses. Iconic memory for vision. Haptic memory for touch. From the wikipedia entry on Sensory memory:

 

 

SM is not involved in higher cognitive functions such as consolidation of memory traces or comparison of information.[18] Likewise, the capacity and duration of SM cannot be influenced by top-down control; a person cannot consciously think or choose what information is stored in SM, or how long it will be stored for.[4] The role of SM is to provide a detailed representation of our entire sensory experience for which relevant pieces of information can be extracted by short-term memory (STM) and processed by working memory (WM).[2] STM is capable of storing information for 10–15 seconds without rehearsal while working memory actively processes, manipulates, and controls the information. Information from STM can then be consolidated into long-term memory where memories can last a lifetime. The transfer of SM to STM is the first step in the Atkinson–Shiffrin memory model which proposes a structure of memory.

[Ralf11]

The research results are quite consistent.

 

 

Sent from my iPhone using Computer Audiophile

 

which ones?

[Jud]

which ones?

 

The past decades of academic research papers regarding human memory for various aspects of sound and what affects the ability to identify those aspects over time.

 

 

Sent from my iPhone using Computer Audiophile

[mmerrill99]

Research into the matter doesn't mean no differences can be discerned longer term. Simply that the fineness of perceptions or in perceiving smallest differences degrades substantially. Not unique to hearing either. You have similar differences related to all your senses. Iconic memory for vision. Haptic memory for touch. From the wikipedia entry on Sensory memory:

 

SM is not involved in higher cognitive functions such as consolidation of memory traces or comparison of information.[18] Likewise, the capacity and duration of SM cannot be influenced by top-down control; a person cannot consciously think or choose what information is stored in SM, or how long it will be stored for.[4] The role of SM is to provide a detailed representation of our entire sensory experience for which relevant pieces of information can be extracted by short-term memory (STM) and processed by working memory (WM).[2] STM is capable of storing information for 10–15 seconds without rehearsal while working memory actively processes, manipulates, and controls the information. Information from STM can then be consolidated into long-term memory where memories can last a lifetime. The transfer of SM to STM is the first step in the Atkinson–Shiffrin memory model which proposes a structure of memory.

 

But it may be that in this area small differences of the type that are only perceivable within the echoic memory timespan, are not that important to our perception of & enjoyment of music? Focus on this type of detail is perhaps the wrong emphasis when it comes to the differences that people report anecdotally - soundstage depth & solidity increasing, air around instruments, better emotional connection to the music/players/singers, etc. Some differences are not conducive to instant A/B differentiation.

[mmerrill99]

There are a number of interesting aspects coming out of this meta-analysis - not that they are new aspects but rather that they re-emphasise what is already established

First this from his conclusion:

" Results showed a small but statistically significant ability of test subjects to discriminate high resolution content, and this effect increased dramatically when test subjects received extensive training"

 

Note the importance of "extensive training". It would be interesting to see just what the statistical result would be if only those studies that engaged in extensive pre-training were included? This is the problem with statistics - a 50% result doesn't necessarily mean that there is no difference - if half the people could hear no difference & half the people could definitely hear a significant difference, the result of this combined would be 50%

I have seen people criticise both of these elements - the small statistical figure & the need for pre-training

 

What these criticisms reveal is an unfamiliarity with perceptual testing & statistical analysis. I don't claim to be an expert in either but I know enough to realise that sensory testing involves testing the perceptions & these are non-exact - so even a fairly obvious differences will be missed some of the time (it may be due to boredom, distraction, second guessing on'e choice, etc.) - the point is that what is being tested is one particular aspect of the workings of a very complex organism & not a machine.

 

I've seen adequate pre-training critiised on the basis that those who claim "night & day differences" should be able to do such a test without pre-training. Again this shows a failure to realise the nature of such tests & the accepted guidelines which recommend adequate training.

 

One other aspect that should be noted - the author states this "One interesting thing that I found, which I did not at all expect, was that most of the potential biases would introduce false negatives. That is, most of the issues were things like not using high res stimuli, or having a filter in the playback chain that removed all high frequency content, or using test methodologies that made it difficult for people to answer correctly even when they heard differences."

 

I actually don't know whey he didn't expect this - it's well known that such tests are biased towards false negatives

[Jud]

But it may be that in this area small differences of the type that are only perceivable within the echoic memory timespan, are not that important to our perception of & enjoyment of music? Focus on this type of detail is perhaps the wrong emphasis when it comes to the differences that people report anecdotally - soundstage depth & solidity increasing, air around instruments, better emotional connection to the music/players/singers, etc. Some differences are not conducive to instant A/B differentiation.

 

You might be quite surprised at what fundamental aspects of a sound humans don't recall accurately within more than about 4 seconds. I'll provide a link or two later to research I linked in another thread.

 

 

Edit: However, I believe you're onto something regarding "focus on this type of detail."

 

 

Sent from my iPhone using Computer Audiophile

[mmerrill99]

You might be quite surprised at what fundamental aspects of a sound humans don't recall accurately within more than about 4 seconds. I'll provide a link or two later to research I linked in another thread.

Sent from my iPhone using Computer Audiophile

Sure but are these "fundamental aspects" of importance to our auditory perception in this hobby where connection to the auditory illusion created by our playback systems? So, are you saying that these "fundamental aspects of sound" that we don;t accurately recall outside our echoic memory timespan, are important "to our perception of & enjoyment of music"?

 

 

We seem to store long-term aspects or features of sound which allows us to immediately identify our family members voices or allows us to immediately recognize & categorise the many different sounds of rain on a roof & differentiate it from the many different sounds of applause or the many different sounds of a crackling fire. What aspects of these sounds we store which allow us this facility to differentiate such diverse spectrum of sound is not fully understood - it is thought that we store some statistical spectrum related to these types of sounds.

[Jud]

Are you saying that these "fundamental aspects of sound" that we don;t accurately recall outside our echoic memory timespan, are important "to our perception of & enjoyment of music"?

Sure but are these "fundamental aspects" of importance to our auditory perception in this hobby where connection to the auditory illusion created by our playback systems? We seem to store long-term aspects or features of sound which allows us to immediately identify our family members voices or allows us to immediately recognize & categorise the many different sounds of rain on a roof & differentiate it from the many different sounds of applause or the many different sounds of a crackling fire. What aspects of these sounds we store which allow us this facility to differentiate such diverse spectrum of sound is not fully understood - it is thought that we store some statistical spectrum related to these types of sounds.

 

Is it important that an audio system play back the correct notes?

 

 

Sent from my iPhone using Computer Audiophile

[mmerrill99]

Is it important that an audio system play back the correct notes?

Sent from my iPhone using Computer Audiophile

 

You have switched the discussion from auditory perception to the device accuracy - why?

 

That's the point & why I asked you the question. If you wish to consider this - would a deep understanding of the weightings that auditory perception pays to various aspects of it's analysis/processing of the waveform impinging on the eardrum have any significance to how our audio playback chain is designed?

 

In other words, if perceptual algorithms worked perfectly (some claim they do) would it change your question?

 

The question I would ask is more along the lines of " What aspects of the sound are of importance to our auditory perception & what are the weightings of these aspects"

 

It seems to me that auditory processing is an afterthought in this hobby - the attention is all placed on the technical accuracy of playback when in fact this may well be a focus on the wrong aspects as ultimately it's our flawed auditory perception that makes the judgement about what we are listening to & not some scope or spectrum anayzer

[Jud]

That's the point & why I asked you the question. If you wish to consider this - would a deep understanding of the weightings that auditory perception pays to various aspects of it's analysis/processing of the waveform impinging on the eardrum have any significance to how our audio playback chain is designed?

 

In other words, if perceptual masking algorithms worked perfectly (some claim they do) would it change your question?

 

The question I would ask is more along the lines of " What aspects of the sound are of importance to our auditory perception & what are the weightings of these aspects"

 

Is it important to you to be able to tell an oboe from a French horn?

 

 

Sent from my iPhone using Computer Audiophile

[mmerrill99]

Is it important to you to be able to tell an oboe from a French horn?

 

Sent from my iPhone using Computer Audiophile

It might be considered more productive in a discussion to deal with points in what I posted but .........?