I made a sample rate blind-test (192 vs 48)

[Scytales]

I read the “medical” study and it’s interesting, thank you. The authors therefore recorded a Gamelan in DSD and noted the presence of significant signals up to 100 Khz.
I couldn't find what model of microphone they used, because the microphones generally used for the production of music records stop much earlier in frequency: you can't have everything, a small capsule going very very high in the treble and a residual noise level intrinsic to the microphone which is very very low... So we use microphones having a larger capsule, rising less high in the treble: like Neuman which cut around 20 Khz, or other brands and models rising higher in the treble, the Bruel and Kjaer for example which peak nevertheless cut at 30 Khz...
On the other hand, in a concert hall the microphones are not placed a few tens of centimeters from the instruments of an orchestra or a solo instrument... and they have little chance of capturing many harmonics located in high frequencies exceeding 20 or 22 Khz... of course if you put a measuring microphone in the mouth of a trumpet or the "bells" of a gamelan... it will be different.
All this to say that super tweeters are cautery on a wooden leg when they are asked to reproduce something that is not on the original tape.

Here is another more recent paper (open publication under Creative Commons License : http://creativecommons.org/licenses/by/4.0/ ) published in Nature by Hiroshi Nittono, one of the authors of the first above mentionned article.

High‑frequency sound components of high‑resolution audio are not detected in auditory sensory memory, Hiroshi Nittono, Nature, 2020.

 

[vintologi]

Seriously , How old were you? 12?

You like the NIH

The Audible Spectrum - Neuroscience - NCBI Bookshelf

It was hard to judge and depended on volume. I was maybe 25 back then.

btw: i just tested again with a 20Khz tone and i could hear it easily when i turned up the volume a little bit (at least with my left ear).

I am like 31 now so i was afraid my hearing had degrading significantly.

Maybe i should use a slightly higher sampling rate (such as 52 KS/s) next time just to make sure i cannot actually hear those frequencies?

 

[vintologi]

Here is another more recent paper (open publication under Creative Common License : http://creativecommons.org/licenses/by/4.0/ ) published in Nature by Hiroshi Nittono, one of the authors of the first above mentionned article.

High‑frequency sound components of high‑resolution audio are not detected in auditory sensory memory, Hiroshi Nittono, Nature, 2020.

Thanks for that paper.

The approach i made when comparing 192 to 48 blind was to see what track i enjoyed the most rather than trying to remember differences in sound. I felt that the experience listening to the 48 version was a bit lacking. You are probably going to fail if you try to focus on minute differences in what you hear directly since it seems like you cannot directly hear those higher frequencies anyway.

And i also had to hunt down special songs that seemed to benefit from ultrasonics (which was few and far between).

 

[Scytales]

I wish to ask you one question.

The audio file you have used and provided a link through (see opening post) is said to have been published on CD in 16 bits/44.1 kHz and downloadable in FLAC 16/44.1. The recording itself has been made in June 1991 : https://cdn.naxosmusiclibrary.com/sharedfiles/booklets/ARU/booklet-ARTS47391-2.pdf (see last page, back cover art of the CD album).

Did you used this source file or CD disc to produce the 48 kHz and 192 kHz sample rate files to do your test ?

If so, the difference spectrum you have publised in your opening post is obviously not that of original ultrasonic content in the original recording, but that of the residual (probably noise shaped quantization noise) produced by the upconversion digital processing to 192 kHz to create one of your files.

 

[vintologi]

I wish to ask you one question.

The audio file you have used and provided a link through (see opening post) is said to have been published on CD in 16 bits/44.1 kHz and downloadable in FLAC 16/44.1. The recording itself has been made in June 1991 : https://cdn.naxosmusiclibrary.com/sharedfiles/booklets/ARU/booklet-ARTS47391-2.pdf (see last page, back cover art of the CD album).

Did you used this source file or CD disc to produce the 48 kHz and 192 kHz sample rate files to do your test ?

If so, the difference spectrum you have publised in your opening post is obviously not that of original ultrasonic content in the original recording, but that of the residual (probably noise shaped quantization noise) produced by the upconversion digital processing to 192 kHz to create one of your files.

Might be a good idea to take a closer look then. Got it from what.cd years ago and it was probably originally from this (now dead) download link:

The New Organ of the Philharmonie Mercatorhalle Duisburg

With this new release a very special highlight of the long-standing partnership between the Duisburg Philharmonics under the baton of Jonathan Darlington and ACOUSENCE has been documented...

www.highresaudio.com

Here is the difference between the files amplified 39dB and slowed down 4x:

difference25%

MediaFire is a simple to use free service that lets you put all your photos, documents, music, and video in a single place so you can access them anywhere and share them everywhere.

www.mediafire.com

Makes me wonder, could it be the case that merely adding ultrasonic noise would achieve the same "hypersonic effect" as accurate ultrasonic reproduction? (assuming the hypersonic effect is real in the first place).

Another problem with that test where i got 9 out of 10 correct was that it played the 48 version first for the first 9 trials which may have skewed the outcome.

 

[Cbdb2]

I remember doing a similar test years ago with this track and i though the 192 version was better (9 times out of 10)

MediaFire

MediaFire is a simple to use free service that lets you put all your photos, documents, music, and video in a single place so you can access them anywhere and share them everywhere.

www.mediafire.com

I used audacity to downsample and upsample (for the 48 KS/s version).

I used https://www.random.org/ integer generator to decide if the 192 version was going to be called A (if 0) or B (if 1).

Album: The New Organ of the Philharmonie Mercatorhalle Duisburg

Song: Thalben-Ball: Variations on a theme of Paganini for pedals

Note that you probably want tweeters that reach well above 30Khz for this.

The difference between the files is definitely not audible by itself (yes i did try listening to it):
View attachment 344912

That does not look like its coming from the music. What instrument produces that much more 70khz than 35khz? The harmonic content of instruments usually decreases with higher harmonics and theres no fundamentals above 10khz.

 

[vintologi]

That does not look like its coming from the music. What instrument produces that much more 70khz than 35khz? The harmonic content of instruments usually decreases with higher harmonics and theres no fundamentals above 10khz.

Probably noise-shaping or some weird noise picked up during recording.

The file look a bit sus in audacity, i should check with abobe audition or something like that.

edit: i may have accidentally exported both as 16 bit instead of 24 bit (both versions), not sure if that would make any difference though.

 

[vintologi]

Here is a link to the 44.1 vs 96 test i did earlier (i liked the sound of the 44.1 version better)

MediaFire

MediaFire is a simple to use free service that lets you put all your photos, documents, music, and video in a single place so you can access them anywhere and share them everywhere.

www.mediafire.com

A and B are my conversions in audacity.

 

[Cbdb2]

Did you know very few mics used to record music pick up anything past 50khz? Most don't make it to 30khz.

 

[dtaylo1066]

A similar test was done a few years back, with 318 participants: https://www.realhd-audio.com/?p=6993. The study has been presented at an AES convention. Outcome:

"Hi-Res Audio or HD-Audio provides no perceptible fidelity improvement over a standard-resolution CD or file. CD-spec and hi-res audio versions sound identical to vast majority of listeners through systems of all kinds".

At age 67 and hearing maybe to 13K at best, I am not sure I could statistically ascertain any difference, nor do I wish to debunk any results.

I do not doubt that the 48 vs. 192 sample rate provided no audible difference to the vast majority of listeners in a test. But what if you or your hearing places you in the minority that may have been able to detect a difference?

We all hear differently, and some listen better or more closely than others. The very, very few Hi-Res tracks I listen to I "believe" sound better to me, but I think the main reason is not sample rate but that they are not massed produced albums and more attention was paid to production.

How about 48K vs lossy?

 

[SIY]

Sigh. The same crank papers that fail replication and are taken seriously by exactly zero researchers not named "Oohashi" or one of his "students" keep coming up again and again with even more fanciful excuses.

Maybe in some biolab, we could make a clone containing Oohashi and Kunchur's DNA and reach Peak Crank.

 

[Geert]

I do not doubt that the 48 vs. 192 sample rate provided no audible difference to the vast majority of listeners in a test. But what if you or your hearing places you in the minority that may have been able to detect a difference?

The few (literally) that could tell a difference were actually not really convinced about their performance: "Most of those that did better than chance observed that it was very difficult to tell them apart and lacked a high degree of confidence". So when listening casually instead of testing it probably wouldn't make any difference. This is the conclusion from the study:

(https://www.realhd-audio.com/?p=7143).

 

[Deleted member 21219]

Speakers are never 100% linear so unless you use a completely separate driver for the ultrasonics there will be intermodulated distrortion from the ultrasonics interacting with the lower frequencies.

Please ... provide a measurement or analysis that supports this. Without that, it is just speculation.

Jim

 

[dtaylo1066]

The few (literally) that could tell a difference were actually not really convinced about their performance: "Most of those that did better than chance observed that it was very difficult to tell them apart and lacked a high degree of confidence". So when listening casually instead of testing it probably wouldn't make any difference. This is the conclusion from the study:

View attachment 345257
(https://www.realhd-audio.com/?p=7143).

Yes, I am aware, and not arguing or wanting to debunk, and how often is music used as background vs critical listening? Then throw in the speaker capability, room, etc. I simply wonder if a small % of listeners with very acute and accurate hearing can denote a difference?

I have suggested a number of times that people on this Forum, which is so dedicated to transparent measurements, which makes total sense to me, should have an in-depth hearing test and evaluation conducted in order to see how good or bad their hearing and perception of sound is. I had a full hearing test done several years ago. Very eye opening and interesting, and one reason why, I believe, subjective audio reviewing is so prevalent. We all hear differently, and it's not flat from 20-20K, along with many other factors.

At some point performance measurement deltas become irrelevant, and the differences imperceptible other than in the measuring device or in the calculus/physics.

 

[Haskil]

Did you know very few mics used to record music pick up anything past 50khz? Most don't make it to 30khz.

And even less than that... if we take the famous Neumann: we stop well before 30 Khz...

 

[krabapple]

j h. c, Oohashi's stuff again?

It never dies.

 

[krabapple]

Here's an example of the gibberish he's published outside of the 'hypersonic effect' fad:

Perspective of the Kansei brain function research

In order to provide a mature definition of Kansei, we introduced a concept from system control theory to arrange the function of the "brain as a system that controls the higher animal behavior". Based on our view of behavior control hierarchy resulting from a brain function, we would like to...

ieeexplore.ieee.org

and here's his attempt to investigate 'hi rez' video the same way he did audio

KANSEI Evaluation on Granularity of Mediated Visual Stimuli. | CiNii Research

With the advance of multimedia technology, the frequency of exposure to artificial visual information has been increasing. The physiological damages caused by visual information which is fewer conformable to human brain has become a serious social problems. We must take into account the fact...

cir.nii.ac.jp

 

[Cbdb2]

KANSEI Evaluation on Granularity of Mediated Visual Stimuli. | CiNii Research

With the advance of multimedia technology, the frequency of exposure to artificial visual information has been increasing. The physiological damages caused by visual information which is fewer conformable to human brain has become a serious social problems. We must take into account the fact...

cir.nii.ac.jp

And after all that work the big conclusion: "Our results suggest that finer visual images (blah blah) give better impression than rougher images." Wow, earth moving discovery.

 

[Cbdb2]

And even less than that... if we take the famous Neumann: we stop well before 30 Khz...

You need small diaphram condensers for ultrasonics. There are a few that are popular in studios. Nueumann KM184 and AKG C451 come to mind (both popular on cymbals). The problem is theres no F response diagrams that go past 20khz. As far as Im concerned if Nueumann and AKG dont think anything past 20khz matters why should I.

 

[staticV3]

You need small diaphram condensers for ultrasonics. There are a few that are popular in studios. Nueumann KM184 and AKG C451 come to mind (both popular on cymbals). The problem is theres no F response diagrams that go past 20khz. As far as Im concerned if Nueumann and AKG dont think anything past 20khz matters why should I.

Or just go with MEMS. $10 and you can record up to ~80kHz.
Plus there's a frequency response graph:

Ultrasonic Mic Board Knowles SPU0410LR5H-QB

Knowles Ultrasonic MEMS mic SPU0410LR5H-QB mounted on a small PCB for easy connections Good usable ultrasonic response up to 80 kHz 3 wire connection20Hz to 10kHz +/- 2 dB10kHz to 80kHz +/- 10 dB For connection details and full specification please see the Knowles datasheet below: KNOWLES...

micbooster.com