[dericchan1]

Folks, there is absolutely no point to even discuss another software in a HQplayer thread. I do not care and do not need to know what the other piece of software does.

 

[camrector]

Folks, there is absolutely no point to even discuss another software in a HQplayer thread

I would agree for a standalone program, but here there is history of these two programs being used together. Knowing how one program might influence the behavior of another is important to know.

 

[dericchan1]

I would agree for a standalone program, but here there is history of these two programs being used together. Knowing how one program might influence the behavior of another is important to know.

Then I would say don’t do it, just stick with using hqplayer

 

[bogi]

I've been thinking about this point. You posted this to demonstrate that since the digitised music signal has infinite bandwith, the dac filter will necessarily have to ring to turn it into a band-limited signal again.

Those spectral images on the above pictures and ringing are not the same thing. Spectral images exist in frequency domain and ringing appears in time domain.

One cannot avoid spectral images (or replicas, that's matter of terminology) of audio band when using digitally sampled signal. Existence of spectral images is the reason why oversampling and analog filtering is used. Oversampling moves the nearest image far from audio band so that DAC analog filter can easier remove these unwanted images.

Digital filtering cannot filter out spectral images. When you apply a digital filter, the filtered result in base band area will get replicated in spectral images too. So the only way to get rid of spectral images is filtering in analog domain.

Ringing is result of band limiting, which is necessary part of resampling. Ringing occurs when a fast transient exceeding Nyquist rate is filtered out. The first usual source of ringing are delta sigma ADCs. They generate high rate sigma delta modulated signal, which si then still internally downsampled (decimated) to lower PCM rate. The second place where downsampling is usually performed is mastering to lower sample rate format. In the case of 44.1k the filter transition band is very narrow (22.05 - 20 = 2.05 kHz) and thus the filter needs to be very steep if we don't want to limit content below 20kHz. Steeper is the filter in frequency domain more it rings in time domain.

The paradoxical thing here is that we have an infinitely ringing perfect filter

I will rather let Jussi to respond this ...

 

[camrector]

My analysis of the situation is that some group of people choose products because of measurements and assume that all the people do the same. That's why we have arguments like: "you think you hear x, because you've been influenced by y". However, for many this is about sound. When I hear about some new toy in town, I get intrigued and try it in my system. If I don't like it, I move on. If there is something I like, then I start to dig deeper on what are the potential variables that are in play as those lead me towards the system that best matches my preferences.

I’m the same way. That digging deeper for me involves many things including MEASUREMENTS to help me understand why I’m hearing what I’m hearing. By seeing how a filter or modulator behaves I can focus on others that perform similar.
I don’t use measurements to tell anyone what’s good or bad. I use measurements to help describe WHY I hear what I’m hearing.

 

[kennyb123]

The point being you admitted to trashing DSD yet were listening to it the whole time. Lol

No I wasn’t. My use of HQPlayer has been for PCM scaling using sinc-M or sinc-L with NOS mode turned on. I only occasionally used it for DSD scaling but never liked what I heard.

Why would misrepresent what I was doing? Rhetorical question. It’s obvious why you would.

You also misrepresented what was posted on the PGGB website by omitting text that doesn’t support your narrative. Unbelievable.

 

[camrector]

No I wasn’t. My use of HQPlayer has been for PCM scaling using sinc-M or sinc-L with NOS mode turned on. I only occasionally used it for DSD scaling but never liked what I heard.

Why would misrepresent what I was doing? Rhetorical question. It’s obvious why you would.

You also misrepresented what was posted on the PGGB website by omitting text that doesn’t support your narrative. Unbelievable.

your dac converts any incoming signal to n multi bit SDM (dsd)
That gets converted to analog.
So yes you were.

Nothing has been misrepresented or taken out of context. Clear posted statements.

 

[plumpudding2]

Those spectral images on the above pictures and ringing are not the same thing. Spectral images exist in frequency domain and ringing appears in time domain.

One cannot avoid spectral images (or replicas, that's matter of terminology) of audio band when using digitally sampled signal. Existence of spectral images is the reason why oversampling and analog filtering is used. Oversampling moves the nearest image far from audio band so that DAC analog filter can easier remove these unwanted images. Analog filter does not introduce ringing - that's about digital domain.

Digital filtering cannot filter out spectral images. When you apply a digital filter, the filtered result in base band area will get replicated in spectral images too. So the only way to get rid of spectral images is filtering in analog domain.

Ringing is result of band limiting, which is necessary part of resampling. Ringing occurs when a fast transient exceeding Nyquist rate is filtered out. The first usual source of ringing are delta sigma ADCs. They generate high rate sigma delta modulated signal, which si then still internally downsampled (decimated) to lower PCM rate. The second place where downsampling is usually performed is mastering to lower sample rate format. In the case of 44.1k the filter transition band is very narrow (22.05 - 20 = 2.05 kHz) and thus the filter needs to be very steep if we don't want to limit content below 20kHz. Steeper is the filter in frequency domain more it rings in time domain.



I will rather let Jussi to respond this ...

Thanks for your reply! I've already asked Jussi this some time ago and he replied that you cannot compare theoretical perfect signals and real world signals in this way if I remember correctly.

On to your other points, a dac uses a reconstruction filter to take the digitized signal and filter out the spectral images so that the baseband signal remains right? It's true that filters only ring when encountering content above their cutoff frequency, but that's exactly what happens when a dac reconstruction filter encounters the spectral images no?
I'd say a dacs reconstruction filter rings by design, but maybe I'm misunderstanding something here..
I thought the defining trait of a Nos dac is they lack such a reconstruction filter.

 

[bogi]

On to your other points, a dac uses a reconstruction filter to take the digitized signal and filter out the spectral images so that the baseband signal remains right?

Yes... D/A stage in the most narrow sense creates signal which already appears in analog domain, but it yet needs to be filtered out by analog reconstruction filter to limit output to area of our interest (audio band). Unfiltered out of band content could influence downstream equipment like class D amplifiers - it potentionally could intermodulate into audio band, or cause some heat or oscilations.

It's true that filters only ring when encountering content above their cutoff frequency, but that's exactly what happens when a dac reconstruction filter encounters the spectral images no?

The term ringing exists in analog domain too, but probably in a different sense than in digital domain, where it is tightly bound to sample rate. No sample rate exists in analog domain. I will leave it to @jlaako to answer your interesting and surprising question.

 

[jlaako]

The paradoxical thing here is that we have an infinitely ringing perfect filter that reconstructs the original signal exactly, it must logically follow that this ringing is part of the reconstruction or in other words adding lots of ringing together gives you the not-ringing original signal.

What looks like "perfect" filter (if we talk about steepness) from frequency domain point of view is absolutely opposite from time domain point of view. Because these are related through inverse 1/x relationship.

Likewise, in mathematically related aspect, when you do spectrogram with FFT (like in HQPlayer metering view) you need to choose transform length. When you make the transform longer, you gain more frequency resolution (in digital filters this means steeper roll-off), but your time resolution to detect changes suffers. You need to choose suitable balance between the two.

This time-frequency uncertainty is called Fourier uncertainty principle. Also remember that hearing can beat this uncertainty principle. See also here.

 

[jlaako]

I'd say a dacs reconstruction filter rings by design, but maybe I'm misunderstanding something here..

Practically yes. And so does ADC decimation filter, so the ringing gets baked into the source data already.

With apodizing filters you can replace/reduce the ringing with less disturbing and possibly shorter one.

Solution is to keep high enough sampling rate throughout, and that's what you get with DSD recordings.

I thought the defining trait of a Nos dac is they lack such a reconstruction filter.

As we know, that will have much more problems by misplacing the transients and in worst case getting transients lost completely. Essentially you'd have tons of transient edge jitter.

 

[Solasuke]

@jlaako

Thank you for your help. I feel as though as I have tried everything and I cannot get HQPlayer to recognise anything over the network. I have bought new microSD cards, I have flashed them with Red OS, RopieeeXL, your HQplayer NAA OS, and nothing is visible in HQPlayer. There is even some difficulty getting them recognised in Roon, though that eventually happens.

I have the Holo red disabled in Roon so it definitely cannot be interefered with. I have tried with the network button set and unset in HQPlayer. I have installed different versions of RopieeeXL. I have flashed the SD cards on a different computer because I was sometimes getting flashing validation errors.

My impression is HQPlayer is somehow blocked from seeing things. I have tried playing with various network settings but nothing seems to make a difference.

Your help is greatly appreciated.

EDIT: I've now tried installing Roon and HQPlayer on another computer and finally managed to see my Holo Red (with an older version of RopieeeXL) over the network. I tried switching back to my main computer and still I see nothing here. I can't see any firewall blocks on my HQPlayer either.

 

[jlaako]

EDIT: I've now tried installing Roon and HQPlayer on another computer and finally managed to see my Holo Red (with an older version of RopieeeXL) over the network. I tried switching back to my main computer and still I see nothing here. I can't see any firewall blocks on my HQPlayer either.

First thing to check is that firewall is not blocking the traffic, and that you have only one network interface active (Control Panel -> Network connections -> Manage network connections). Main reason for not finding a NAA is that outbound multicast packets end up on wrong interface. They should usually go where your default route is (internet connection). But sometimes with multiple active network interfaces the multicast discovery packets end up in wrong interface. Another alternative is that the response is not getting back due to firewall.

 

[Solasuke]

First thing to check is that firewall is not blocking the traffic, and that you have only one network interface active (Control Panel -> Network connections -> Manage network connections). Main reason for not finding a NAA is that outbound multicast packets end up on wrong interface. They should usually go where your default route is (internet connection). But sometimes with multiple active network interfaces the multicast discovery packets end up in wrong interface. Another alternative is that the response is not getting back due to firewall.

Hooray! Thanks for your advice, the outbound packets look to have been routing to the Local Area Connection instead of Wifi. Disabling the Local Area Connection fixed it!

 

[jlaako]

Hooray! Thanks for your advice, the outbound packets look to have been routing to the Local Area Connection instead of Wifi. Disabling the Local Area Connection fixed it!

Great! If you are using WiFi, the new NAA v5 protocol may be beneficial compared to the older v4 one. NAA OS provides v5 while I believe Red OS provides v4. I don't know what RopieeeXL provides. This mostly as heads up in case you encounter dropouts or something like that.