[chesebert]

It does not “need” upsampling. All these HQplayer games are just another trendy feature that is fun to play with.

Digital stores like NativeDSD often sell music recorded in hi-res both in the original and in a lower bitrate (for those with old DACs). Anyone can buy an album in a few bitrates and compare the difference in sound to what a HQplayer do to a lower bitrate with it’s upsampling.

The result of upsampling sounds interesting, but has nothing in common with a sound of a real high-res recording. HQplayer changes the sound, some may like it, but to say that upsampling is supposedly the "correct" way to listen to music is simply not true.

I think you are confused. We are talking about moving aliased images due to not having a reconstruction filter to a much higher frequency range. This has nothing to do with high res recording.

 

[chesebert]

Holo products are sold in EU with CE certificate issued in the name of Holo Technology Co. Ltd. I can’t confirm company registration because PRC commercial register is in Chinese language, but certificate is issued by a well known service, so everything should be legit.

I don’t know how things are in USA, but in Europe the seller is obliged to provide a guarantee. The user is not required to interact with the manufacturer at all。

I couldn't find Holo's business license or any trace of the company registration in China. It's a Chinese company so US practice is not relevant.

 

[Ragnar-BY]

I think you are confused. We are talking about moving aliased images due to not having a reconstruction filter to a much higher frequency range.

I prefer listening to music, not to “aliased images”

 

[chesebert]

I prefer listening to music, not to “aliased images”

You are by definition listening to "aliased images" if you use a NOS dac. People love their music with a smidge of reflection; go figure.

 

[sajunky]

You are by definition listening to "aliased images" if you use a NOS dac. People love their music with a smidge of reflection; go figure.

Aliases are created during A/D conversion if there is energy above Nyquist. This process is irreversible. During D/A conversion there is no aliases, as there is no energy above Nyquist, so nothing to be aliased.

NOS DAC create images. Images do not affect sound in the acoustic band as long your downstream channel is free of distortions. Distorted images leak to the audio band, then is a problem.

Images are continuously filtered out on the downstream equipment, so on speakers where distortions are big, do not play significant role. In fact, most of NOS DACs do not have high order analogue low pass filter. But a quality of your amplification is important (distortion free up to 100kHz), no question about.

 

[Riccioli30]

I prefer listening to music, not to “aliased images”

Yea,I've been hearing these technical discussions since the 80s and the digital that I've chosen over time has always and only chosen by listening.

 

[chesebert]

Aliases are created during A/D conversion if there is energy above Nyquist. This process is irreversible. During D/A conversion there is no aliases, as there is no energy above Nyquist, so nothing to be aliased.

NOS DAC create images. Images do not affect sound in the acoustic band as long your downstream channel is free of distortions. Distorted images leak to the audio band, then is a problem.

Images are continuously filtered out on the downstream equipment, so on speakers where distortions are big, do not play significant role. In fact, most of NOS DACs do not have high order analogue low pass filter. But a quality of your amplification is important (distortion free up to 100kHz), no question about.

I think you are talking about under sampling, which is not what I am saying. 44khz sampling is simply not high enough to not require a reconstruction filter. I believe in practice you will have aliasing issues unless you resample to to a much higher frequency. At the end of the day, NOS method is a compromised reproduction of the original audio signal. If you like the "sound" then so be it.

 

[paradoxper]

Yea,I've been hearing these technical discussions since the 80s and the digital that I've chosen over time has always and only chosen by listening.

That's a long time. I'd be bored of the tired and beaten argument.

 

[sajunky]

I think you are talking about under sampling, which is not what I am saying. 44khz sampling is simply not high enough to not require a reconstruction filter. I believe in practice you will have aliasing issues unless you resample to to a much higher frequency. At the end of the day, NOS method is a compromised reproduction of the original audio signal. If you like the "sound" then so be it.

Incorrect, I am positive. There is no aliasing on decoding as there is no energy above Nyquist. Images coexist with sound without interfering as long there are not distorted as there are in different bands. If distorted, intermodulation products leak to the audio band.

In addition, there is no problem with 3dB NOS drop at 20kHz, it doesn't need to be compensated. It is measured, but not perceived. A reason is that ultrasonic energy (while not heard as a sound) stimulate sensors increasing sensitivity of top frequencies. Our hearing is not analogue, but consist of thousands digital sensors, in case you didn't know.

 

[chesebert]

Incorrect, I am positive. There is no aliasing on decoding as there is no energy above Nyquist. Images coexist with sound without interfering as long there are not distorted as there are in different bands. If distorted, intermodulation products leak to the audio band.

In addition, there is no problem with 3dB NOS drop at 20kHz, it doesn't need to be compensated. It is measured, but not perceived. A reason is that ultrasonic energy (while not heard as a sound) stimulate sensors increasing sensitivity of top frequencies. Our hearing is not analogue, but consist of thousands digital sensors, in case you didn't know.

I don't know what you mean by "energy". When you do DTFT you get image at 2fs, 3fs, etc. Nothing is perfect and you get some reflection around the fs unless your fs is a large number.

Why are we talking about sampling? Just buy whatever that sounds good to you...it's not rocket science.

 

[sajunky]

I don't know what you mean by "energy". When you do DTFT you get image at 2fs, 3fs, etc. Nothing is perfect and you get some reflection around the fs unless your fs is a large number.

Everything is based on energy. Matter is energy, so you are. Your atoms, protons, electrons... whathever, interract each other. But there are rules. By example electrons on different energy levels do not interfere each other.

No energy no influences. 'Reflections' are called images. FFT plot shows you density of energy per frequency. If there is energy there is influence, but only with components of the same frequency. Everything is like in ideal world. There are rules is this world where there is no time coordinate, but frequency and energy density. When analysing FFT plot, need to understand these rules, there are different than in our world. Energy is a major property in this world. And you are asking what is energy....

 

[adrianm]

Majority of the manufacturer dont even have high end analyser to confirm good design. Just because dcs or msb costs an arm and a leg doesnt make them magical?

dCS actually claim they created their own tools for testing because what's out there isn't accurate enough. Or magical enough, however you want to put it

 

[GoldenSound]

dCS actually claim they created their own tools for testing because what's out there isn't accurate enough. Or magical enough, however you want to put it

Have they given much info on this?
Whilst developing in house tools is not uncommon, this is usually either for cost saving reasons, sometimes a particular bit of test gear can be exorbitantly priced and in an engineering focused firm making a similar tool in-house may actually be cheaper, but this won't be the case for developing a substitute for a totl scope or analyzer though.

OR

For testing something specific to your use case where tools don't already exist, usually software related. In this instance I'd imagine that what they're referring to is simulations for the DAC's mapper. Noise shapers and things related to that are usually simulated because they're purely mathematical with completely predictable results, meaning developments and decisions can be simulated, iterated on and developed far quicker than needing to test in the analog domain. Also when testing at the output of the DAC, random analog noise obscures quantization noise and it's basically impossible to separate, making analysis of noise shaper performance below the analog noise floor essentially impossible and therefore you have to do it in software.

A manufacturer actually showed me some of their own simulations for a very advanced noise shaper recently, can't give details cause NDA, but it was quite interesting how in-depth that can get.

 

[adrianm]

Have they given much info on this?
Whilst developing in house tools is not uncommon, this is usually either for cost saving reasons, sometimes a particular bit of test gear can be exorbitantly priced and in an engineering focused firm making a similar tool in-house may actually be cheaper, but this won't be the case for developing a substitute for a totl scope or analyzer though.

OR

For testing something specific to your use case where tools don't already exist, usually software related. In this instance I'd imagine that what they're referring to is simulations for the DAC's mapper. Noise shapers and things related to that are usually simulated because they're purely mathematical with completely predictable results, meaning developments and decisions can be simulated, iterated on and developed far quicker than needing to test in the analog domain. Also when testing at the output of the DAC, random analog noise obscures quantization noise and it's basically impossible to separate, making analysis of noise shaper performance below the analog noise floor essentially impossible and therefore you have to do it in software.

A manufacturer actually showed me some of their own simulations for a very advanced noise shaper recently, can't give details cause NDA, but it was quite interesting how in-depth that can get.

I remember i read it as part of the Rossini Apex review in Stereophile:

“The challenge dCS faced as it developed the Ring DAC Apex, Hales said, was the inability of conventional test equipment to measure the Apex's performance values with sufficient accuracy. In an email, dCS VP of Sales and Marketing John Giolas wrote, "the linearity of our DACs is so [much higher than] the industry norm, we've had to create our own test equipment to measure it. Conventional test equipment is wholly inadequate to measure what our DACs are capable of.”

 

[piaseczek]

dCS actually claim they created their own tools for testing because what's out there isn't accurate enough. Or magical enough, however you want to put it

They mastered the marketing, it's so simple.

Brutal reality shows that there is 6dB difference in SINAD. It's a lot.
https://goldensound.audio/2021/10/12/dcs-bartok-with-headphone-amp-measurements/