Can we trust our ears?

[Julf]

Genelec posted an article on their blog with some nice pictures on applying research on"slow listening" to pro-audio monitoring.

Written by marketing people?

 

[audiophile]

Written by marketing people?

Written by Thomas Lund, Genelec’s Senior Technologist

From http://www.aes.org/events/147/presenters/?ID=8706:
Thomas Lund has been active in pro audio research and development since 1995, primarily in the areas true-peak detection, sound exposure, localization, sense of space and loudness metering. From a background in medicine and perception, Thomas has taken part in audio standardization globally, and written more than 50 academic papers.

Thomas Lund is current chairman of a TC108X/WG3 under the European Commission tasked with preventing hearing loss from spreading because of over-exposure to Personal Media Players; and he holds position as Senior Technologist at Genelec R&D.

 

[Blumlein 88]

Written by marketing people?

You had it right the first time. Slow listening makes no sense.

 

[HemiRick]

I just saw something that proved what we hear is influenced by what we see. At least when it comes to speech. A person was making the lip motion for a "B" vs an "F", but the sound played was exactly the same. What you heard depended on the lip motion.

 

[Julf]

I just saw something that proved what we hear is influenced by what we see. At least when it comes to speech. A person was making the lip motion for a "B" vs an "F", but the sound played was exactly the same. What you heard depended on the lip motion.

http://mentalfloss.com/article/72587/mcgurk-effect-or-brains-are-weird

 

[Juhazi]

And Audioholics published this article a week ago. Good reference list!
https://www.audioholics.com/room-acoustics/mind-over-music

 

[tuga]

I haven't read the whole thread, hope this hasn't been posted before:

 

[Thomas Lund]

A review generally doesn't include new research. This one discusses theories (e.g. active sensing, perceptual bandwidth, latency of sensation etc.) in the light of a number of new in-vivo studies, and relate them to pro audio. The few "odd" references are there as examples of precisely that, along with 100 relevant ones.

The misrepresentation here of perceptual bandwidth - that seeing has the highest by far - confuses physiological and perceptual bandwidth. Other main take-aways from the review should be: Active sensing, efferent pathways to the ears are as important as afferent ones, movement is essential in hearing, cross-modal influence should not be under-estimated, adults mainly hear what they expect to hear.

There is little doubt most books on human perception will need revision over the coming years, so we're actually merely suggesting a research agenda based on evidence from other fields of science.

To get facts instead of rumours, please read the paper, https://tonmeistertagung.com/download/tmt30-2018-proceedings.pdf pg 111ff.

PS. For the record, I have never postulated humans can hear ultrasonic (>20 kHz) sound. On the contrary, actually. However, in pro monitoring, there can be acoustic consequences of ultrasonic sonic sound being reproduced vs (steep) 22 kHz filtering.

 

[anmpr1]

Active sensing ...movement is essential in hearing, cross-modal influence should not be under-estimated...

This reminded me of two humorous (but serious) comments about driver alignment on a loudspeaker baffle--first from the late Drew Daniels, JBL applications engineer, regarding his 'home made' large footprint low distortion loudspeaker:

There is bound to be "time-smearing" or "image-smearing" from any sound source that is not a simple point in space, but by aligning the system elements in a straight vertical line (except the subwoofer drivers), horizontal time and image smearing is eliminated. Humans don't perceive vertical time and image smearing unless they jump up and down in front of the speaker system--a practice I don't recommend for critical listening (divides your attention).

And the time Arnie Nudell complained to Peter Aczel that the LS3/5a could not be considered a 'point source', at which point Aczel replied:

As for the Rogers, you're just trying to hassle us. You know as well as we do that the few inches between the two drivers of the LS3 shrink to virtually nothing--i.e. a point, from the acoustic perspective of a listener ten or more feet away. Of course to a very short listener a very short distance from the speaker it might not appear as a point source.

 

[Thomas Lund]

This reminded me of two humorous (but serious) comments about driver alignment on a loudspeaker baffle--first from the late Drew Daniels, JBL applications engineer, regarding his 'home made' large footprint low distortion loudspeaker

Good points in relation to this thread. A well designed "point source" also has the advantage of making (vertical plane) reflections coherent.

I'm actually using a very nearfield setup for headphone comparisons as we write, see picture. Standing on the heart provides a 50 cm listening triangle (relatively) away from boundaries. The elephants between the speakers is the price I have to pay for doing this in our living room

 

[krabapple]

A review generally doesn't include new research. This one discusses theories (e.g. active sensing, perceptual bandwidth, latency of sensation etc.) in the light of a number of new in-vivo studies, and relate them to pro audio. The few "odd" references are there as examples of precisely that, along with 100 relevant ones.

The misrepresentation here of perceptual bandwidth - that seeing has the highest by far - confuses physiological and perceptual bandwidth. Other main take-aways from the review should be: Active sensing, efferent pathways to the ears are as important as afferent ones, movement is essential in hearing, cross-modal influence should not be under-estimated, adults mainly hear what they expect to hear.

There is little doubt most books on human perception will need revision over the coming years, so we're actually merely suggesting a research agenda based on evidence from other fields of science.

To get facts instead of rumours, please read the paper, https://tonmeistertagung.com/download/tmt30-2018-proceedings.pdf pg 111ff.

PS. For the record, I have never postulated humans can hear ultrasonic (>20 kHz) sound. On the contrary, actually. However, in pro monitoring, there can be acoustic consequences of ultrasonic sonic sound being reproduced vs (steep) 22 kHz filtering.

As in

"For investigation of audio questions of possible long-term influence, we should consider all four quadrants of Fig. 2, including both interior and exterior aspects over time, but especially when examining any quality which may not be completely familiar.
.
.
.

In case what is tested for is unfamiliar, slow listening could take as long as it would for the subject to learn a new language, maybe more. "

Yes, sure, fine.

But in the real world of the audio enthusiasts, one constantly encounters the listener who implicitly claims to be so completely familiar with many 'qualities' one would expect to be subtle at best if not inaudible, that they can immediately say that change X has caused A to sound different from, and better or worse than, B.

And sometimes, in their parlance 'even my spouse could hear it'.

Audio, outside of a lab setting, revolves around claims like *those*. From hobbyists, influencers, audio journalists, engineers, who claim to 'trust their ears' (though they are using their eyes and memories too). Which gets us back to the title of this thread.

Your paper will be waved around as proof by *those* people


What is to be done?

 

[anmpr1]

But in the real world of the audio enthusiasts, one constantly encounters the listener who implicitly claims to be so completely familiar with many 'qualities' one would expect to be subtle at best if not inaudible, that they can immediately say that change X has caused A to sound different from, and better or worse than, B.

And sometimes, in their parlance 'even my spouse could hear it'.

The wife thing is a common joke. While rummaging through some historical material I think I found the first instance of it. Or at least an early instance. The old Vacuum Tube Valley magazine [Issue 14, 2000]. An interview with David Hafler, certainly no tweako subjectivist. Dave was describing the development of the Ultralinear circuit with his partner Herb Keroes at Acrosound, before Dynaco:

We didn't have distortion meters at that time because we couldn't afford them on the scale of business we were doing. I looked at what [Herb] had done and asked, "Why not use negative feedback with a separate winding?, it shouldn't be any worse and maybe it will be better."

I strapped [the transformer] up and put the thing into the amp I had at home. The phone rang and I went to answer it. I had just put down the stylus on the record... my wife came running in to me and said, "What did you do?" It was the new transformer using negative feedback on the extra winding. Herb said, "Well, maybe that's it!"

 

[Thomas Lund]

What is to be done?

Agreed, which is why the review also says "objective criteria, not immune to falsification, and repeatable procedures need to be established so “more testing time” does not become a way of defending just any claim." New in-vivo measurements could become objective anchors, if we actively search for them, for instance using ERPs, fMRI, MEG, fNIRS - or even invasive probing.

We partly wrote the review out of wonder why lossy data reduction sounded so awful. Its artefacts are everywhere, in streaming, on radio, in the subway, on the plane, even in precious national archives. Didn't the codec test subjects hear it back then like we do now? Maybe they were more sensitive to hiss, wow&flutter, loss of HF, azimuth phasing etc.

ITU-R BS.1116 and BS.1534 are excellent tools but, they have let us down every now and then.

 

[pozz]

Jim Austin seems to be more of an objectivist to me. He prefers transparent sounding DACs and once famously double-blind tested dCS Bartok DAC (yes, DBT, in Stereophile with levels matched !) The manufacturer wasn't happy about it ))


His article summarizes two AES papers by Genelec researchers:

Time for Slow Listening
http://www.aes.org/e-lib/browse.cfm?elib=20547

On Human Perceptual Bandwidth and Slow Listening
http://www.aes.org/e-lib/browse.cfm?elib=19621

These papers basically claim that short ABX tests are inaccurate since humans can analyze maximum 40-50 bits of audio data per second.

https://www.audiosciencereview.com/...n-stoddard-on-blind-testing.11716/post-627043

Linking to my reply in the thread above for reference.

 

[Thomas Lund]

An important academic initiative that will likely cause many a textbook on hearing/perception to be rewritten,
https://intelligentsoundengineering.wordpress.com/2021/01/14/aural-diversity/

 

[Julf]

An important academic initiative that will likely cause many a textbook on hearing/perception to be rewritten,
https://intelligentsoundengineering.wordpress.com/2021/01/14/aural-diversity/

Maybe, maybe not. We will see. I do see that it is sponsored by the Arts & Humanities Research Council, not anything related to psychology, neurology, physics or perception research.

 

[pma]

An important academic initiative that will likely cause many a textbook on hearing/perception to be rewritten,
https://intelligentsoundengineering.wordpress.com/2021/01/14/aural-diversity/

I agree that "The assumption that we all possess a standard, undifferentiated pair of ears underpins most listening scenarios" is wrong and it can be proven by different results of different individuals when they perform the same ABX listening test. It can be seen in many studies that different people from the test group had different results and also here in the ABX tests posted. Ear + brain evaluation is not the simple engineering circuit, even if many oriented to engineering only would wish so.

 

[Julf]

I agree that "The assumption that we all possess a standard, undifferentiated pair of ears underpins most listening scenarios" is wrong and it can be proven by different results of different individuals when they perform the same ABX listening test. It can be seen in many studies that different people from the test group had different results and also here in the ABX tests posted. Ear + brain evaluation is not the simple engineering circuit, even if many oriented to engineering only would wish so.

Of course. But that different hearing affects the person no matter if the sound is reproduced from a recording or produced live. The human mind is good at compensating for physical shortcomings. It doesn't imply that the system that reproduces recorded sound should be different for different people.

 

[Leporello]

^Exactly.

 

[anmpr1]

An important academic initiative that will likely cause many a textbook on hearing/perception to be rewritten...

The notion that 'we all hear differently' has to be understood correctly, and inferences made accordingly. With pretty much any bio function, whether it is temperature, heart rate, reaction time to stimulus, visual acuity, or hearing, there is going to be variance around a mean. The components of audiological investigation-- FR, speech discrimination, speech reception, pure tone, air and bone conduction will vary, but that doesn't mean that the stimulus itself can be questioned. And it doesn't mean that, on a perceptual level, outliers above the mean are able to 'hear' something unique, that someone with comparable acuity will not be able to hear.

As far as audio reproduction in hi-fi goes, the basics of sound recording and recreation, FR, distortion types, etc., are well known and are not waiting to be discovered. In any case, the idea that 'we all hear differently' has nothing to do with whether one can actually prove that they hear what they claim to hear.

The linked researcher's goal is to find avenues to improve the hearing of those with auditory impairments (loss, tinnitus etc). That goal is worthy and anything they do to help folks with hearing loss is welcome.