[GoldenSound]

Did I ask directly from you that why nos is special ? Still instead of replying you came with more questions ? Why are you irritated when I ask about your nos affinity without any objective reason ? Now please don't report me as you did earlier without any reason just for asking straight technical reasons why nos is better or why nos at all should exist, while there is no technical superiority you could prove. Since you started asking me instead of replying, I again request you to answer my simple question why as per you nos is superior ?

I never said nos has any technical superiority.
I said that some may subjectively prefer it. Calm down

Though the argument for time domain accuracy given the lack of pre/post ringing could be made.

You constantly assert that nos has poor measurements despite the fact that repeatedly in this thread measurements have been linked which disprove that.

And as to reporting your messages, I didn't. Perhaps your trolling is annoying others

 

[rkt31]

I never said nos has any technical superiority.
I said that some may subjectively prefer it. Calm down

Though the argument for time domain accuracy given the lack of pre/post ringing could be made.

You constantly assert that nos has poor measurements despite the fact that repeatedly in this thread measurements have been linked which disprove that.

And as to reporting your messages, I didn't. Perhaps your trolling is annoying others

Ok then please post sweep test of cd quality nos, if it does not have poor measurements. Btw do you have any technical proof that nos has better time domain characteristics when you don't have even full frequency spectrum available due to treble roll off ?

 

[GoldenSound]

Ok then please post sweep test of cd quality nos, if it does not have poor measurements. Btw do you have any technical proof that nos has better time domain characteristics when you don't have even full frequency spectrum available due to treble roll off ?

What tests are you wanting to see?
I've already linked several

https://www.superbestaudiofriends.org/index.php?threads/kte-may-technical-measurements.8933/

Perhaps rather than constantly dodging questions and throwing out random new assertions you should look at the content that has already been linked previously in this thread.
I would hope you don't require spoon feeding

 

[rkt31]

What tests are you wanting to see?
I've already linked several

https://www.superbestaudiofriends.org/index.php?threads/kte-may-technical-measurements.8933/

Perhaps rather than constantly dodging questions and throwing out random new assertions you should look at the content that has already been linked previously in this thread.
I would hope you don't require spoon feeding

Do you call these simple,straight and interrelated questions random ? (1) what is preringing artifacts and how they look like in sound wave ( not in a single impulse ) (2) how nos retains time domain accuracy when it can't even retain the faithful frequency response (3) how nos is technically superior and special as compared to os (4) if bigger stair steps of nos are not problem then why the whole audiophile world is going mad about hqplayer and pggb and using with dacs which are nos or known for it's nos (read holo may ). I found your reviews and post more technical than others so it does not look good when instead of answering these simple questions you try to divert the matter by throwing unrelated counter arguments.

 

[ShenaRingo326]

Ares 2 is a lovely DAC especially for the price. At the moment it and the D1SE are my recommendations at that pricepoint depending on user preference.
There are only two things I don't like about the Ares 2:

1) The Delay makes it a no-go for movies/games. But of course not an issue for music
2) Denafrips 'NOS' is not NOS. It's linear interpolation oversampling, so it's somewhat misleading for them to call it NOS.

But it sounds great

Did you prefer OS + Slow, OS + Fast, or NOS with HQPlayer on the ARES II?

 

[GoldenSound]

Do you call these simple,straight and interrelated questions random ? (1) what is preringing artifacts and how they look like in sound wave ( not in a single impulse ) (2) how nos retains time domain accuracy when it can't even retain the faithful frequency response (3) how nos is technically superior and special as compared to os (4) if bigger stair steps of nos are not problem then why the whole audiophile world is going mad about hqplayer and pggb and using with dacs which are nos or known for it's nos (read holo may ). I found your reviews and post more technical than others so it does not look good when instead of answering these simple questions you try to divert the matter by throwing unrelated counter arguments.

I'll answer your questions, hopefully you'll answer mine too.

1) Exactly as described. Here is some actual music and the result using two filters to show how there is a clear difference in ringing between two filters with real music:

Here is some additional work done by Archimago to show both phase shifting and ringing introduced by digital filters:

2) Because there is no phase shift introduced and no pre/post ringing. Whilst also preserving sample values which something like spline or polynomial interpolation (or MQA) would not.
Frequency and time domain behaviour are two completely separate things. Hence why we measure and test them separately. You can't test time domain consistency or jitter from FR either for example.

3) I never said it was, please stop making this assertion. My view is that filters are always a tradeoff and suit different hardware and types of content. Having the flexibility to choose and change is the best solution imo.

4) Stair steps are caused by stuff far outside the audible band (44.1khz). Whereas ringing occurs at the nyquist frequency (22.05khz). Meaning that either neither of these are an issue because they're outside the audible band. Or both are but in different ways (and with NOS arguably being less problematic cause no human can possibly hear 44.1khz whereas they might be able to sliightly hear 22.05khz).
You cannot consider one a problem but not the other. They are both ultrasonic issues.
You can only see these two issues when viewing the analog out recording at a very high sample rate. If you were to record a nos dac sine at 44.1khz you wouldn't see any stair steps at all.

Holo may 44.1khz sample rate 1khz sine:

People (including myself) like HQPlayer because it's about as good as you can get for oversampling. But it's not 'Better' than NOS.
If you ARE going to oversample then using proper noise shapers and going to as high a rate as possible is ideal.
But I switch between NOS and OS because they're simply different. Neither is perfect.

Now, if you wouldn't mind answering my question, that would be appreciated:

Why do you keep jumping in to trash anything when the word 'NOS' is mentioned, and why do you keep asserting that NOS dacs have poor measurements when that isn't the case at all as demonstrated several times?
If you really really like chord dacs or something then that's totally fine. But jumping into threads to stir drama and mock things which you seemingly don't understand is childish and frustrating for all involved.

 

[gdwallasign]

I never said nos has any technical superiority.
I said that some may subjectively prefer it. Calm down

Though the argument for time domain accuracy given the lack of pre/post ringing could be made.

You constantly assert that nos has poor measurements despite the fact that repeatedly in this thread measurements have been linked which disprove that.

And as to reporting your messages, I didn't. Perhaps your trolling is annoying others

I ignored him but didn't report... Maybe that was it

 

[rkt31]

I'll answer your questions, hopefully you'll answer mine too.

1) Exactly as described. Here is some actual music and the result using two filters to show how there is a clear difference in ringing between two filters with real music:



Here is some additional work done by Archimago to show both phase shifting and ringing introduced by digital filters:


2) Because there is no phase shift introduced and no pre/post ringing. Whilst also preserving sample values which something like spline or polynomial interpolation (or MQA) would not.
Frequency and time domain behaviour are two completely separate things. Hence why we measure and test them separately. You can't test time domain consistency or jitter from FR either for example.

3) I never said it was, please stop making this assertion. My view is that filters are always a tradeoff and suit different hardware and types of content. Having the flexibility to choose and change is the best solution imo.

4) Stair steps are caused by stuff far outside the audible band (44.1khz). Whereas ringing occurs at the nyquist frequency (22.05khz). Meaning that either neither of these are an issue because they're outside the audible band. Or both are but in different ways (and with NOS arguably being less problematic cause no human can possibly hear 44.1khz whereas they might be able to sliightly hear 22.05khz).
You cannot consider one a problem but not the other. They are both ultrasonic issues.
You can only see these two issues when viewing the analog out recording at a very high sample rate. If you were to record a nos dac sine at 44.1khz you wouldn't see any stair steps at all.

Holo may 44.1khz sample rate 1khz sine:


People (including myself) like HQPlayer because it's about as good as you can get for oversampling. But it's not 'Better' than NOS.
If you ARE going to oversample then using proper noise shapers and going to as high a rate as possible is ideal.
But I switch between NOS and OS because they're simply different. Neither is perfect.

Now, if you wouldn't mind answering my question, that would be appreciated:

Why do you keep jumping in to trash anything when the word 'NOS' is mentioned, and why do you keep asserting that NOS dacs have poor measurements when that isn't the case at all as demonstrated several times?
If you really really like chord dacs or something then that's totally fine. But jumping into threads to stir drama and mock things which you seemingly don't understand is childish and frustrating for all involved.

Exactly that's what I wanted you to post here and you did. See only sinc can recreate the original band width limited signal exactly, everything else is approximation. The issue is large number of coefficient you use closer to sinc function, more accurate the wave form will be. Since you can't have infinite long process, the resulting wave will always be slightly different than original. Now the real part, since nos simply use a horizontal line joining the two points by holding the previous value, it is by far the "most deviated" from the original wave and gives the biggest error. "That's why it is the crudest form of os". Even that 32x os linear interpolation by joining two points by a straight line is much better than nos horizontal line in this regard as it is lot closer to original wave than a simple horizontal line. No wonder as you fit the best curve you get closer to original wave and differences from original wave reduce but yes you can't recreate the original wave back due to infinitely long process. But obviously a "long horizontal line of nos itself is the biggest deviation" from original wave. Btw it is getting difficult for me to believe that how a technical person like you can still cling to accuracy of nos.

 

[GoldenSound]

See only sinc can recreate the original band width limited signal exactly, everything else is approximation

This is entirely dependent on the ADC filter

Since you can't have infinite long process, the resulting wave will always be slightly different than original

Yes, one of those issues being pre-ringing as shown in the images.

it is by far the most deviated from the original wave

This is simply mathematically false and shows a misunderstanding of how sampling theorem works.

Even that 32x os linear interpolation by joining two points by a straight line is much better than nos horizontal line in this regard as it is lot closer to original wave

No it isn't. The original wave would have all sorts of high frequency content which we cannot predict. It wouldn't look anything like any of these images.
We are re-constructing a BAND LIMITED signal here, not the full original thing.

Btw it is getting difficult for me to believe that how a technical person like you can still cling to accuracy of nos.

Unfortunately if you don't understand the difference between an in and out of band signal and how they relate to filter behaviour it's going to be quite tricky to explain it to you.

Also, are you going to answer my question now?

 

[rkt31]

I'll answer your questions, hopefully you'll answer mine too.

1) Exactly as described. Here is some actual music and the result using two filters to show how there is a clear difference in ringing between two filters with real music:



Here is some additional work done by Archimago to show both phase shifting and ringing introduced by digital filters:


2) Because there is no phase shift introduced and no pre/post ringing. Whilst also preserving sample values which something like spline or polynomial interpolation (or MQA) would not.
Frequency and time domain behaviour are two completely separate things. Hence why we measure and test them separately. You can't test time domain consistency or jitter from FR either for example.

3) I never said it was, please stop making this assertion. My view is that filters are always a tradeoff and suit different hardware and types of content. Having the flexibility to choose and change is the best solution imo.

4) Stair steps are caused by stuff far outside the audible band (44.1khz). Whereas ringing occurs at the nyquist frequency (22.05khz). Meaning that either neither of these are an issue because they're outside the audible band. Or both are but in different ways (and with NOS arguably being less problematic cause no human can possibly hear 44.1khz whereas they might be able to sliightly hear 22.05khz).
You cannot consider one a problem but not the other. They are both ultrasonic issues.
You can only see these two issues when viewing the analog out recording at a very high sample rate. If you were to record a nos dac sine at 44.1khz you wouldn't see any stair steps at all.

Holo may 44.1khz sample rate 1khz sine:


People (including myself) like HQPlayer because it's about as good as you can get for oversampling. But it's not 'Better' than NOS.
If you ARE going to oversample then using proper noise shapers and going to as high a rate as possible is ideal.
But I switch between NOS and OS because they're simply different. Neither is perfect.

Now, if you wouldn't mind answering my question, that would be appreciated:

Why do you keep jumping in to trash anything when the word 'NOS' is mentioned, and why do you keep asserting that NOS dacs have poor measurements when that isn't the case at all as demonstrated several times?
If you really really like chord dacs or something then that's totally fine. But jumping into threads to stir drama and mock things which you seemingly don't understand is childish and frustrating for all involved.

Exactly that's what I wanted you to post here and you did. See only sinc can recreate the original band width limited signal exactly, everything else is approximation. The issue is larger coefficient you use closer to sinc function the more accurate the wave form will be. Since you can't have infinite long process, the resulting wave will always be slightly different than original. Now the real part since nos simply use a horizontal line joining the two points by holding the previous value it is by far the most deviated from the original wave. Even that 32x os linear interpolation by joining two points by a straight line is much better than nos horizontal line in this regard as it is lot closer to original wave than a simple horizontal line. No wonder as you fits the best curve you get closer to original wave and differences from original wave reduces but yes you can't recreate the original wave back due to infinitely long process. But obviously a big horizontal line of nos itself is the biggest deviations from original wave. Btw it is getting difficult for me to believe that how a technical person like can still cling to accuracy of nos

This is entirely dependent on the ADC filter



Yes, one of those issues being pre-ringing as shown in the images.



This is simply mathematically false and shows a misunderstanding of how sampling theorem works.



No it isn't. The original wave would have all sorts of high frequency content which we cannot predict. It wouldn't look anything like any of these images.
We are re-constructing a BAND LIMITED signal here, not the full original thing.



Unfortunately if you don't understand the difference between an in and out of band signal and how they relate to filter behaviour it's going to be quite tricky to explain it to you.

Also, are you going to answer my question now?

So you mean the horizontal line in nos which joins two points retains the original samples ? The values lying on horizontal line are far far away from original wave. In fact nos introduces all incorrect points between two original points. Now please no more tangential replies.

 

[GoldenSound]

Exactly that's what I wanted you to post here and you did. See only sinc can recreate the original band width limited signal exactly, everything else is approximation. The issue is larger coefficient you use closer to sinc function the more accurate the wave form will be. Since you can't have infinite long process, the resulting wave will always be slightly different than original. Now the real part since nos simply use a horizontal line joining the two points by holding the previous value it is by far the most deviated from the original wave. Even that 32x os linear interpolation by joining two points by a straight line is much better than nos horizontal line in this regard as it is lot closer to original wave than a simple horizontal line. No wonder as you fits the best curve you get closer to original wave and differences from original wave reduces but yes you can't recreate the original wave back due to infinitely long process. But obviously a big horizontal line of nos itself is the biggest deviations from original wave. Btw it is getting difficult for me to believe that how a technical person like can still cling to accuracy of nos

So you mean the horizontal line in nos which joins two points retains the original samples ? The values lying on horizontal line are far far away from original wave. In fact nos introduces all incorrect points between two original points. Now please no more tangential replies.

You do understand what band limiting is right?

Original:

20khz band-limited and reconstructed with sinc interpolation:

Do they look the same?
No. The original cannot be reconstructed. That's just not how it works. You can reconstruct upto nyquist only. Nothing more.

 

[rkt31]

You do understand what band limiting is right?

Original:

20khz band-limited and reconstructed with sinc interpolation:


Do they look the same?
No. The original cannot be reconstructed. That's just not how it works. You can reconstruct upto nyquist only. Nothing more

You do understand what band limiting is right?

Original:

20khz band-limited and reconstructed with sinc interpolation:


Do they look the same?
No. The original cannot be reconstructed. That's just not how it works. You can reconstruct upto nyquist only. Nothing more.

That's what sinc interpolation does, recreate the original bandwidth limited signal. Good to see atleast you agreed to something.

 

[GoldenSound]

That's what sinc interpolation does, recreate the original bandwidth limited signal. Good to see atleast you agreed to something.

So you DO understand what band-limited means?
And yet you keep going on about 44.1khz stuff?

 

[GoldenSound]

I'm going to leave this here until you answer even one of the questions I asked.
I've made several long posts answering your questions and having what I think is a considerable amount of patience @rkt31

It's only fair you answer the ones I've asked you.

If not, well, I'd encourage you to change your mindset. Everyone on the forum benefits when topics are approached with the intent to discuss and learn.
When you disrupt threads with random assertions and childish emoji filled posts which refuse to answer anything put to you all it does is frustrate people and make us wish that you'd crawl back under your bridge so we could continue having a civil discussion

 

[rkt31]

So you DO understand what band-limited means?
And yet you keep going on about 44.1khz stuff?

44.1khz can theoretically have information upto 22.05khz but you can't accurately get that information without sinc interpolation. Everything else will not be accurate, specially yours nos horizontal line which introduces points between two samples which are far far away from original. See any line joining two samples is os, you can't simply have individual peaks equivalent to digital samples, you have to fill the gap and nos horizontal line is the most inaccurate.