[theveterans]

@castleofargh

I'll provide what I think/believe/opinion is the disconnection between the posts from sunjam and to you guys as I think the concept of digital signal processing on a basic level isn't fully grasped by @sunjam yet, and for that reason sunjam hasn't interpreted of Monty's talking points to the math/science behind it in a "non cherrypicked" manner. For the record, I do not believe sunjam is manipulating me or influencing what I know about the math and behavior of DACs and how they translate sampling to a perfect (in terms of audibility thresholds) sinusoidal wave. It's just that if I put my shoes to you and gregorio, I can understand where you guys are coming from (sunjam cherrypicking and stuff), but in my opinion, sunjam just hasn't had the full grasp of the concept yet and why would these variety of custom filters that influence the final DAC output (on 44.1 KHz original sample rate so long as reconstructed analog output provides almost 0.907 at Nyquist, yes that's inaudible (0.1 dB delta) from the ideal 1.0, and Sinc-L 16 million tap filter would probably be around 0.990 at Nyquist) be so frigging minute (in terms of audibility thresholds) that can explain why 99.9% of audiophiles fail a properly set DBT ABX volume matched test consistently.

I already provided my sighted listening biases on my previous posts earlier, but ultimately succumb to the power of placebo when put to the golden standard ABX test

Given my thinking style, I asked them for the proof that why they believe objectively that "Hi-Res is useless".

They mainly come back with five reasons:

1. Sampling theory said 44.1k is enough
2. Based on science, the difference is so samll so that you cannot hear it
3. They show me the Monty's video and it proves that 44.1k can reconstruct perfect smooth sine wave ouput
4. You are not bat, you cannot hear higher than 22k
5. There are research reports to show that you cannot hear the difference

In honesty, these 5 reasons though are 100% supported by math. However, when applied in real life without any controls, the combination of brain/mood/sight/hearing/feeling can make us perceive the difference on a subconscious level. Delving into evolution theory, these traits does make sense to our species survival. Full disclosure, I embrace the biases though since this makes the audio hobby incredibly fun and not boring

 

[sunjam]

Just noticed these beauties!

That’s because it didn’t, it just proved that 1bit DSD is not suitable for high quality audio applications. You know, the format you think is great but isn’t! That’s funny.

Ah... no wonder HQPlayer is so good in DSD upsampling as I heard that HQPlayer uses multi-level SDM (I need to confirm later about it).

=======

By the way, what is your view about this interesting fact:

Ultrasonic vibrations of 100 kHz or over applied directly to the skull (i.e. mastoid) are known to produce auditory sensation through bone conduction and resonance. They are not felt as vibratory or thermal sensations, but rather as normal tones with a lower frequency. Similar to auditory stimulation, they produce brain evoked responses and mismatch detection responses to frequency diference and stimulus omission, suggesting that the common auditory pathway is used for bone-conducted ultrasound and air-conducted normal sound perception.

It is from the paper you mentioned in an early reply below

I have not seen any verified, reliable evidence that anyone can, even under 10 year olds. There is quite lot which indicates that adults can’t though. This paper by Nittono in 2020, ...

 

[sunjam]

@castleofargh

I'll provide what I think/believe/opinion is the disconnection between the posts from sunjam and to you guys as I think the concept of digital signal processing on a basic level isn't fully grasped by @sunjam yet, and for that reason sunjam hasn't interpreted of Monty's talking points to the math/science behind it in a "non cherrypicked" manner. For the record, I do not believe sunjam is manipulating me or influencing what I know about the math and behavior of DACs and how they translate sampling to a perfect (in terms of audibility thresholds) sinusoidal wave. It's just that if I put my shoes to you and gregorio, I can understand where you guys are coming from (sunjam cherrypicking and stuff), but in my opinion, sunjam just hasn't had the full grasp of the concept yet and why would these variety of custom filters that influence the final DAC output (on 44.1 KHz original sample rate so long as reconstructed analog output provides almost 0.907 at Nyquist, yes that's inaudible (0.1 dB delta) from the ideal 1.0, and Sinc-L 16 million tap filter would probably be around 0.990 at Nyquist) be so frigging minute (in terms of audibility thresholds) that can explain why 99.9% of audiophiles fail a properly set DBT ABX volume matched test consistently.

I already provided my sighted listening biases on my previous posts earlier, but ultimately succumb to the power of placebo when put to the golden standard ABX test


In honesty, these 5 reasons though are 100% supported by math. However, when applied in real life without any controls, the combination of brain/mood/sight/hearing/feeling can make us perceive the difference on a subconscious level. Delving into evolution theory, these traits does make sense to our species survival. Full disclosure, I embrace the biases though since this makes the audio hobby incredibly fun and not boring

Cool, thanks for your comment. I'd love to learn from other people what I missed. I need someone to point out my blindspots.

Cheers

 

[sunjam]

@castleofargh

I'll provide what I think/believe/opinion is the disconnection between the posts from sunjam and to you guys as I think the concept of digital signal processing on a basic level isn't fully grasped by @sunjam yet, and for that reason sunjam hasn't interpreted of Monty's talking points to the math/science behind it in a "non cherrypicked" manner. For the record, I do not believe sunjam is manipulating me or influencing what I know about the math and behavior of DACs and how they translate sampling to a perfect (in terms of audibility thresholds) sinusoidal wave. It's just that if I put my shoes to you and gregorio, I can understand where you guys are coming from (sunjam cherrypicking and stuff), but in my opinion, sunjam just hasn't had the full grasp of the concept yet and why would these variety of custom filters that influence the final DAC output (on 44.1 KHz original sample rate so long as reconstructed analog output provides almost 0.907 at Nyquist, yes that's inaudible (0.1 dB delta) from the ideal 1.0, and Sinc-L 16 million tap filter would probably be around 0.990 at Nyquist) be so frigging minute (in terms of audibility thresholds) that can explain why 99.9% of audiophiles fail a properly set DBT ABX volume matched test consistently.

I already provided my sighted listening biases on my previous posts earlier, but ultimately succumb to the power of placebo when put to the golden standard ABX test


In honesty, these 5 reasons though are 100% supported by math. However, when applied in real life without any controls, the combination of brain/mood/sight/hearing/feeling can make us perceive the difference on a subconscious level. Delving into evolution theory, these traits does make sense to our species survival. Full disclosure, I embrace the biases though since this makes the audio hobby incredibly fun and not boring

By the way, just wondering if you want to have a look of my little test about confirmation bias? Mind to share your view about it? confirmation bias test

 

[gregorio]

Ah... no wonder HQPlayer is so good in DSD upsampling as I heard that HQPlayer uses multi-level SDM (I need to confirm with later about it).

If that is the case (that it is not 1bit DSD) then it’s not necessarily inferior to PCM. In other words, it would be just as useless as hi-res PCM as a music distribution/playback format rather than even more useless!

By the way, what is your view about this interesting fact:

That it’s not particularly interesting, as it has nothing to do with music/sound recordings or distribution, or with the consumer reproduction and playback of them.

I'd love to learn from other people what I missed. I need someone to point out my blindspots. … [Plus] just wondering if you want to have a look of my little test about confirmation bias?

In response to ALL these, I remind you that you’ve already demonstrated you’re lying!

G

 

[sunjam]

That it’s not particularly interesting, as it has nothing to do with music/sound recordings or distribution, or with the consumer reproduction and playback of them.

Hmm... I thought this one is directly related to hearing. No?

Does auditory sensation mean hearing?

Do we see the same paragraph as below? Did I miss something again? Or did I mis-interpret or mis-represent anything here?

Ultrasonic vibrations of 100 kHz or over applied directly to the skull (i.e. mastoid) are known to produce auditory sensation through bone conduction and resonance. They are not felt as vibratory or thermal sensations, but rather as normal tones with a lower frequency. Similar to auditory stimulation, they produce brain evoked responses and mismatch detection responses to frequency diference and stimulus omission, suggesting that the common auditory pathway is used for bone-conducted ultrasound and air-conducted normal sound perception.

=================

Ah one more thing, if you don't mind, could you share your view about the difference below?

 

[bfreedma]

Hmm... I thought this one is directly related to hearing. No?

Does auditory sensation mean hearing?

Do we see the same paragraph as below? Did I miss something again? Or did I mis-interpret or mis-represent anything here?


=================

Ah one more thing, if you don't mind, could you share your about the difference below?

Suggestion: Stop trying to immediately respond to every post with attempts to find contradictions/minutia and invest that time in actually reading and absorbing the available research and content posted here. That would make for far better discussion going forward.

It feels very much like you’re using words/ concepts without a baseline understanding of them. While I don’t think you’re an AI driven bot, your responses remind me of those generated by Machine Learning solutions with an insufficiently trained or biased Large Language Model.

 

[gregorio]

It's just that if I put my shoes to you and gregorio, I can understand where you guys are coming from (sunjam cherrypicking and stuff), but in my opinion, sunjam just hasn't had the full grasp of the concept yet and why would these variety of custom filters that influence the final DAC output …

We started from the position of either ignorance, “he hadn’t fully grasped the concept yet” or that he did grasp it but was deliberately trying to mislead others. Almost everything he’s posted since indicates the latter, falsehoods (including fake or misappropriated images) contradictory/incompatible claims, misinterpretations, endless cherrypicking and far more besides. And then, just repeating the same BS even after it’s been explained to him, while deflecting or ignoring the questions and refutations put to him. It’s virtually impossible now to rationally conclude anything other than the latter.

Hmm... I thought this one is directly related to hearing. No?
Do we see the same paragraph as below? Did I miss something again? Or did I mis-interpret or mis-represent anything here?

In the same order as your questions, the answers are: No, Yes, Yes, Yes. We did see the same paragraph but what “ultrasonic vibrations of 100kHz or over” are in music recordings to start with and what audiophiles reproduce music recordings by bolting audiophile speakers “directly to the skull (IE. Mastoid)”? So No, it is not “directly related to hearing” and YES, you “missed something again”, although how you can miss something that you yourself bolded is a mystery (but not really)! lol

your responses remind me of those generated by Machine Learning solutions with an improperly trained or biased Large Language Model.

Now that’s interesting, his responses do indeed remind me of a dodgy AI chatbot!

G

 

[VNandor]

Something that seemingly everyone missed so far is that @sunjam's experience is not even about high-res. He took a 16bit 44kHz sampling rate file, upsampled it to a higher rate (using a 300$ software I might add) and zero padded it to 32bit, then played back this upsampled version of the file with a DAC that does the conversion with a terrible accuracy by design. A higher sampling rate isn't useless for a NOS DAC because it adds at least a little bit of OS to it, although not anywhere near as much as what would be ideal. This doesn't require a 192/24bit file, as the upsampling can be done on a 16bit 44kHz file just as well. Oversampling is used in any normal DAC because it is actually useful in reconstructing the signal better but the original sample rate doesn't matter as the oversampled rate ends up at least in the 1-10MHz range anyways. This doesn't mean the file should be at this sample rate, converting a digital signal to this rate is just a useful intermediate step in analog signal reconstruction. It's also a step that NOS DACs lack.

Upsampling doesn't magically make the file hi-res. Upsampling can't produce any new information out of thin air, all it does is calculating what the analog signal is at certain points in time between the samples with precision that is arbitrarily close to perfect. With a "perfect" filter it can calculate the samples with perfect precision, with anything less, it just degrades the signal a bit (and increases the sample rate of course, and again, a high sample rate is a necessary step for a high quality reconstruction).
A perfect filter is also only required if you want perfect reconstruction right up to the Nyquist frequency. If the filter starts to roll off at 20kHz and reaches the stop band attenuation at the nyquist frequency, then the signal is perfectly reconstructed right up to 20kHz. If it only reached to stopband attenuation at 24kHz, it still would perfectly reconstruct the signal to 20kHz but there would also be some images between 22kHz and 24kHz.

Arguing that the "monty video" is wrong due to the fact that NOS DACs benefit from upsampling (which is again, different from an actual "Hi-Res" file) is like calling someone that claims people have 2 hands a liar by pointing to some unfortunate person that was born with 3 hands instead. It's technically correct that not all people has 3 hands but it doesn't mean that saying people have 2 hands is the same as lying and being misleading or deceptive.

 

[theveterans]

Let's talk about confirmation bias. I think it should be related to our discussion.

How many of you truely understand what is confirmation bias? It is defined as below (source: wiki)



Let's do a very simple test here to see if you really know what is confirmation bias:


If you see 4 different boxes with gradient of greyscales in the middle column, you just say what you see.

What if I told you that there are something wrong with your vision, as you SHOULD see 4 different static grey color boxes instead of gradients.

The following picture show you the second box is actually a static grey color box (this is the actual capture of the second box in the above picture, you can try to capture it and see for yourself):




You may say, "Cool, now, I know I got tricked by my visual perception".

AFTER knowing the fact that all the 4 boxes in the middle column of the first picture are indeed just static grey boxes (instead of gradient of greyscale)..

Now, let me ask you a tough question.... a really really tough question:




What do you see in the middle column of the above picture?

Answer A: 4 different boxes of gradient of greyscale? or
Answer B: 4 different boxes of static grey color?

It is not a trick question.

Just answer it based on what you see.

Your answer may be

"Well....hmm...well... hmm... Answer A...

Hang on.... should I pick B ?"

Should you pick B?

I just ask you what you see.

If you answer B, your answer is affected by confirmation bias.

If you answer A, your answer is just reporting what you can see with the above picture.

(Actual answer: it is exaclty the same picture you saw at the beginning of this page. It is just scaled down to a smaller size)

Do you have confirmation bias? Or you have a different definition of confirmation bias from the wiki I shown earlier?

I answered A: I report what I see (maybe interpreted as biases too as well if one argues that our minds can trick you through illusions). Just look at my posts outside of sound science forum: I pretty much express my opinion in audio setup as in the camp of everything matters: cables, interconnects, DAPs/DACs, amps, power conditioning (i.e. 0.000001V DC ripple filtering/conditioning), but I digress.

In terms of the 5 points I quoted, using common sense of audibility thresholds and understanding of the math behind it, these 5 statements are absolutely true in that point of view with the assumption (which is viewed as common sense here obviously) that you must use a proper reconstruction filter:

sunjam said:
Given my thinking style, I asked them for the proof that why they believe objectively that "Hi-Res is useless".

They mainly come back with five reasons:

1. Sampling theory said 44.1k is enough
2. Based on science, the difference is so samll so that you cannot hear it
3. They show me the Monty's video and it proves that 44.1k can reconstruct perfect smooth sine wave ouput
4. You are not bat, you cannot hear higher than 22k
5. There are research reports to show that you cannot hear the difference

Here's where sound science guys see you as cherry picking information:

The stairstep time domain output of the F5 filter (aka AKM Super-Slow Roll-off) or in a true NOS and filterless DAC: the zero hold order sampling

Audio science guys are pointing out (paraphrasing) that Monty sees no value with streaming higher than 44.1KHz due to DACs having a proper reconstruction filter internally applied to the original sample, and thus the resulting analog signal is silky smooth, FR domain is linear by audibility threshold and quantization noise is inaudible (~90dB SINAD or better). Now, accepting the concept of audibility thresholds (roughly below -70dB of distortion is inaudible due to auditory masking, etc.) will tell you as an audio science guy that the analog signal is well with the boundaries of accurate reproduction unless you diasgree with established audibility thresholds and prove consistently by DBT ABX test that you can distinguished between digital filters using a SOTA DAC such as the Holo Audio May used by goldensound!

Now, from your previous posts, you stated these below:

His intention for creating the video:

He wrote an ariticle claiming "192/24 doesn't make sense". Many people read that and disagreed with his view point. One of the major areas for the argument was the stair step waveform.

Then he created the video to show (aka 'proof') that he is right especially regarding the perfect smooth sine wave reconstruction.

Pseudo science technique:

Probably he wanted to ensure that people would agree what he said, it looks to me (with my trained eyes) he was using some "pseudo science" technique to sell his demos. i.e. using very strong words (e.g. certainly, losslessly, exact, etc) and avoiding or omitting tiny but very important details and creating some illusion. For example,

he emphasized on his video that "a digital waveform is not a stair-step and you certainly don't get stair-step when you convert from digital back to analog" ("fact 1")
he emphasized on his article that "The analog signal can be reconstructed losslessly, smoothly, and with the exact timing of the original analog signal" ("fact 2")

"fact 1" is not correct as the article I used for my blog showed stairstep waveform output from a modern DAC.
"fact 2" is correct only if the reconstruction is under ideal situation.

For "fact 1", he avoided mentioning any thing related to internal oversampling, filtering, NOS DAC, etc...
For "fact 2", he avoided mentioning about the ideal situation.

He use a smooth looking sine wave to demonstrate "fact 1" and "fact 2" are factual in order to create the illusion that his claims are correct.

I believe that it is one of the root cause of the claim "Hi-Res is useless". I feel the obligation to help to debunk these as many people are mis-led by the video and they belief in the claim wholeheartedly.

The counter argument with using fact 1 boils down to audio science guys as "cherry picking" because the zero hold order sampling is broken by design as it does not fullfill linearity within the threshold of audibility (~3 dB down at Nyquist). The counter argument for Fact 2 is that by accepting that there's no such thing as perfect reconstruction filter, but manufacturers like Chord, Schiit, and software developer like HQPlayer push the filter complexity to ridiculous levels (as math shows that anything that is met by low pass filter must have some quantization noise) combined with audibility thresholds (~90 dB SINAD), you shouldn't hear a difference at all from the original analog waveform thus Monty emphasizing, "The analog signal can be reconstructed losslessly, smoothly, and with the exact timing of the original analog signal".

Hope this helps you understand the audio science guys point.

As for me, my sighted biased listening perception, going back to my answer as A: I report what I see/experience/hear, is all over the head-fi site, but if I'm subjected to DBT ABX volume matched test, I lose all that ability to hear the magic staircase waveform sound from my NOS filterless DAC hehe

 

[bfreedma]

Apologies, a little OT but think it’s tangentially relevant to this conversation:

An example of how information sources that the ML/AI solution developers choose to target can influence machine learning responses.

I suspect ChatGPT has its training pointed at the general internet with thousands of audiophile sights vs. only a few audio science based sources. The following was using the older 3.5 Large Language Model. I’ll run the same through ChatGPT’s 4.0 LLM later to see if the current model “does better”. Bottom line - AI/ML solutions are only as good as there discovery/data accumulation models and the specific wording of the user’s question and at this time, should not be assumed to be authoritative.

Q asked:
Is high resolution audio beneficial to listeners vs. CD quality?

ChatGPT 3.5 response:
High-resolution audio can offer benefits to listeners compared to CD quality, especially for those with high-end audio equipment and discerning ears. It provides greater detail, depth, and dynamic range, potentially enhancing the listening experience. However, the difference may not be significant for all listeners or all types of music, so it depends on individual preferences and equipment quality.

Sigh. Thank you TAS, Stereophile, cable vendors, and audiophile sites. Yet another reminder to not use any single source when looking for accurate information.

 

[castleofargh]

@castleofargh

I'll provide what I think/believe/opinion is the disconnection between the posts from sunjam and to you guys as I think the concept of digital signal processing on a basic level isn't fully grasped by @sunjam yet, and for that reason sunjam hasn't interpreted of Monty's talking points to the math/science behind it in a "non cherrypicked" manner. For the record, I do not believe sunjam is manipulating me or influencing what I know about the math and behavior of DACs and how they translate sampling to a perfect (in terms of audibility thresholds) sinusoidal wave. It's just that if I put my shoes to you and gregorio, I can understand where you guys are coming from (sunjam cherrypicking and stuff), but in my opinion, sunjam just hasn't had the full grasp of the concept yet and why would these variety of custom filters that influence the final DAC output (on 44.1 KHz original sample rate so long as reconstructed analog output provides almost 0.907 at Nyquist, yes that's inaudible (0.1 dB delta) from the ideal 1.0, and Sinc-L 16 million tap filter would probably be around 0.990 at Nyquist) be so frigging minute (in terms of audibility thresholds) that can explain why 99.9% of audiophiles fail a properly set DBT ABX volume matched test consistently.

I already provided my sighted listening biases on my previous posts earlier, but ultimately succumb to the power of placebo when put to the golden standard ABX test


In honesty, these 5 reasons though are 100% supported by math. However, when applied in real life without any controls, the combination of brain/mood/sight/hearing/feeling can make us perceive the difference on a subconscious level. Delving into evolution theory, these traits does make sense to our species survival. Full disclosure, I embrace the biases though since this makes the audio hobby incredibly fun and not boring

I did what I should and then more, by now I've read most of his posts(time I'll never get back) his article, the one he took the screenshots from, Monty's article and watched the video(it had been a minute). I did it several times for each by now(to fact-check a bunch of things). Trying to understand, trying to explain, trying to find what seemed unlikely and several times just wasn't true. And I 110% agree with:

We started from the position of either ignorance, “he hadn’t fully grasped the concept yet” or that he did grasp it but was deliberately trying to mislead others. Almost everything he’s posted since indicates the latter, falsehoods (including fake or misappropriated images) contradictory/incompatible claims, misinterpretations, endless cherrypicking and far more besides. A then, just repeating the same BS even after it’s been explained to him, while deflecting or ignoring the questions and refutations put to him. It’s virtually impossible now to rationally conclude anything other than the latter.

But maybe the biggest issue for me is that he presents himself like this:

I was trained in identifying pseudo science claims. I hope I can help to debunk Pseudo science claims as much as I can.

If he was 1/10th of that, he would have died from shame by now. No I'm wrong, he would have never ever written that article the way he did in the first place.
I keep thinking "double standard" without posting it because it doesn't feel like enough to characterize how divorced his pretend position of integrity/debunking is from his actual posts.

But I admit I've been feeling strange from the get-go, even wondering if it was an experiment, and I still can't put my finger on it. To the point that for once, I don't instantly push aside the idea of AI being involved like I usually do when it's brought up. That's how bizarre this is.

 

[gregorio]

Something that seemingly everyone missed so far is …

To be fair, there was some discussion about the difference between hi-res and oversampling much earlier in the thread but I don’t recall where.

Q asked:
Is high resolution audio beneficial to listeners vs. CD quality?

Ask pretty much any “audiophile” question and you get audiophile BS answers. I hope they sort out the learning model, etc. It would be interesting to see what 4.0 comes up with but I won’t hold my breath that it can come up with the truth yet, I think we’re still a long way from that unfortunately.

But maybe the biggest issue for me is that he presents himself like this:

I was trained in identifying pseudo science claims. I hope I can help to debunk Pseudo science claims as much as I can.

If he was 1/10th of that, he would have died from shame by now. No I'm wrong, he would have never ever written that article the way he did in the first place.
I keep thinking "double standard" without posting it because it doesn't feel like enough to characterize how divorced his pretend position of integrity/debunking is from his actual posts.
But I admit I've been feeling strange from the get-go, even wondering if it was an experiment, and I still can't put my finger on it. To the point that for once, I don't instantly push aside the idea of AI being involved like I usually do when it's brought up. That's how bizarre this is.

Hitting “Like” just isn’t a sufficient response to this!

G

 

[bfreedma]

I did what I should and then more, by now I've read most of his posts(time I'll never get back) his article, the one he took the screenshots from, Monty's article and watched the video(it had been a minute). I did it several times for each by now(to fact-check a bunch of things). Trying to understand, trying to explain, trying to find what seemed unlikely and several times just wasn't true. And I 110% agree with:


But maybe the biggest issue for me is that he presents himself like this:

If he was 1/10th of that, he would have died from shame by now. No I'm wrong, he would have never ever written that article the way he did in the first place.
I keep thinking "double standard" without posting it because it doesn't feel like enough to characterize how divorced his pretend position of integrity/debunking is from his actual posts.

But I admit I've been feeling strange from the get-go, even wondering if it was an experiment, and I still can't put my finger on it. To the point that for once, I don't instantly push aside the idea of AI being involved like I usually do when it's brought up. That's how bizarre this is.

There are certainly some indicators that the poster is utilizing a machine learning solution (the vast majority of what is being called AI is actually ML) but the heavy formatting makes me suspect that the OP is using ML to generate responses that he then edits to fit his desired position and adds color to highlight elements of the ML generated text he believes supports that position. Not saying that ML can’t be instructed to format text as seen here, but that takes some knowledge and effort to achieve.

 

[Ryokan]

Maybe he was just practising his English?