[taipan254]

I tried looking into HQ player and PGGB but these upscalers are so unnavigable I'd rather just buy something like an M Scaler so I have the freedom to use my own player...(groove music lol) unless there's another way to play and search my files without opening them from HQplayer? I'd love that but I'm not sure how. It won't even open up root folders of which I have all my music organized into subfolders, so that's a killer. For reference I only listen to FLACs I have downloaded or ripped.

Earlier today, it dawned on me that Foobar2000 may be an option for me, you, and others that want to sample PGGB. There is a "component" / "add-on" to Foobar2000 called PGGB-RT. It is limited in a number of ways compared to PGGB 256 (the offline remastering tool). However, it does allow one to get 64-bit processing of 2bn taps of PGGB for free (1bn better than HQPlayer's Sinc-M filter) and 90 seconds per track of free PGGB at up to 1bn taps (depending on track length).

Today, I added the PGGB-RT component from the Foobar component library. Since I have a streamer, I also downloaded the UPnP Media Renderer output component as well. I then set my ultraRendu streamer on MPD / DLNA mode. Next, I played a few songs with native Gustard filters (i.e., PGGP-RT was INACTIVE) to acclimatize to Foobar. Lastly, I played the same music through Roon via HQP and NAA with no upsampling (no filters, no dither, straight pass-through). This is where I came to conclusion one:

WITHOUT ANY UPSAMPLING, I have to say that Foobar2000, for me, sounds a lot better streaming to my ultraRendu via UPnP than Roon streaming to HQP via NAA. I previously came to the conclusion that Roon streaming to HQP via NAA was better than Roon streaming to RoonBridge via RAAT. I'm bad with audiophile jargon but again I come back to "clarity" / "lack of fuzziness". Foobar2000 + UPnP sounds the "cleanest" to me.​

​

I then went back to Foobar2000. I selected the same music and right clicked resample with PGGB. I also turned on NOS-Mode on the Gustard. I then played all the tracks, and then went back to Roon with HQP via NAA with Sinc-M and LNS15 filters enabled. I did this with a few times with different taps selected in PGGB-RT's settings in Foobar (controlled in the settings tab). I came to conclusion two:

EVEN WITH 1 MILLION TAPS, I have to say that Foobar2000 + PGGB-RT streamed to my ultraRendu via UPnP sounds a lot better than Roon + HQP streamed to my ultraRendu via NAA. Further, with more taps, the more I perceived "reverb" / "texture" of guitars (say the guitar heavy opening of Radiohead's "The Bends") and "decay" / "space" (say the symbols at the start of Danzig's "Last Ride"). The improvements increased as taps increased, albeit at a smaller and smaller scale for each incremental increase in taps. ​

​

Caveats:

1. My laptop sucks: 8 Gigs of RAM, Quadcore Mobile i7855U CPU... My laptop was unable to multitask even with 1mn taps, and the delay to start a new track was substantial. There were also lots of dropouts, especially as the number of taps increased. I would bet a powerful desktop with lots of RAM (32GB) would sound much better and upsample with PGGB-RT much quicker. I also think this explains the diminishing returns I heard with increased taps.
2. My headphones are not "top-tier"... again, I use ThieAudio Oracle Mk1s and a DCA x Drop Aeon Open Back in balanced mode. What has enabled me to perceive differences in upsampling has been the acquisition of an ultraRendu streamer and some power hacks (mains filter, new cheapo audiophile fuses, and ground boxes). However, I definitely think my headphones are also holding me back.
3. Different software configurations sound different. I conducted this comparison in the above manner to control for SQ differences in different software music players. I think it is fairly well-known on other forums (AudiophileStyle, What's Best Forum) that, with respect to sound quality, Roon leaves a lot to be desired relative to lighter-weight players (I personally can confirm Foobar2000 and Audirvana sound better than Roon, and many say Taiko's native player bests everything else on the market, though I have not heard it). I don't know if folks here on head-fi have the same opinion.
4. None of this, except for HQP, works with streaming. The only million+ tap solution for streaming services like Qobuz right now is sadly via HQPlayer. It's better than nothing, but for streaming-centric PCM-based upscalers, having one option sucks.... Maybe PGGB comes up with something better than HQP for real-time playing that is also Roon and end-point compatible. Maybe a higher bit, lower tap real-time processing engine (or a lower bit, higher tap real-time processing engine)! Something far less resource-intensive but sounds better than HQP for PCM upscaling. That seems like smart R&D if I was Zaphod (PGGB's developer). HQP is DSD-focused anyways! PCM upscaling is an afterthought for HQP.
5. All systems are different, as is perception and hearing.

In any case, going back to Roon with HQP and NAA was a "Let Down" (pardon the intended Radiohead pun). I can't stand drop-outs / skipping and I simply cannot multitask with Foobar while PGGB-RT is enabled. Time to build a hefty CPU for processing. I think this Gustard x26 Pro is far more capable than I'm realizing. I'm definitely not getting the most out of it. But I am getting more out of it.

​

 

[kennyb123]

Different software configurations sound different. I conducted this comparison in the above manner to control for SQ differences in different software music players. I think it is fairly well-known on other forums (AudiophileStyle, What's Best Forum) that, with respect to sound quality, Roon leaves a lot to be desired relative to lighter-weight players (I personally can confirm Foobar2000 and Audirvana sound better than Roon, and many say Taiko's native player bests everything else on the market, though I have not heard it). I don't know if folks here on head-fi have the same opinion.

So many good nuggets in your awesome post. The above is spot on. I hear this too so my use of Roon is limited to only when I’m looking to discover unfamiliar music. Squeeze is my preferred choice for playing PGGB’d files. On my Antipodes platform it’s easily the best sounding option.

Roon tends to keep itself too busy and all that activity can lead to noise that is detrimental to sound quality. The lower the activity, the better the sound quality.

 

[Mista Lova Lova]

(I have posted about the amplifier aspect of this question in the Burson Soloist 3XP thread)

I have just discovered that putting my Burson amp in the bypass mode results in a more natural, less-coloured sound signature/presentation. It is surely due to the op-amps being bypassed and only the Class A side of the amp being engaged (unless the bypass mode does not bypass the V6 Vivid op-amps).

However, this means I need to control the volume in a different way. I would not use Roon/HQPlayer solely for this purpose as I'd be afraid that a glitch would make me go deaf (plus my DAC "pops" when switching from PCM to DSD and those pops would be at full volume).

I have, for the first time since owning the Gustard X26 Pro, tried to control the volume using the DAC and I must say my usually-quite-sensitive ears are not detecting an obvious deterioration in sound quality. Am I right in thinking that controlling the volume in this way before it enters the amp in bypass mode would not be detrimental to sound quality, given that the DAC, even in "NOS" mode, controls the volume anyway (as part of its sound processing)?

 

[kennyb123]

I have, for the first time since owning the Gustard X26 Pro, tried to control the volume using the DAC and I must say my usually-quite-sensitive ears are not detecting an obvious deterioration in sound quality. Am I right in thinking that controlling the volume in this way before it enters the amp in bypass mode would not be detrimental to sound quality, given that the DAC, even in "NOS" mode, controls the volume anyway (as part of its sound processing)?

This is how I’ve been using it and, like you, I hear nothing amiss. I hear no loss of goodness at lower volume levels, which is where poorly-implemented volume controls tend to suffer losses. I am guessing that the volume control provided by the ESS 9038 PRO is leverage. The X26pro had an excellent output stage with it’s own dedicated linear supply so that is contributing to the great sound quality.

 

[Mista Lova Lova]

So many good nuggets in your awesome post. The above is spot on. I hear this too so my use of Roon is limited to only when I’m looking to discover unfamiliar music. Squeeze is my preferred choice for playing PGGB’d files. On my Antipodes platform it’s easily the best sounding option.

Roon tends to keep itself too busy and all that activity can lead to noise that is detrimental to sound quality. The lower the activity, the better the sound quality.

Your posts have encouraged me to experiment with Foobar and PGGB. I have been streaming music for a long time, but thankfully I did have some FLAC albums on my hard drive.

To make things equal, what I've done is I've downloaded the ASIO plug-in for Foobar to ensure that it bypasses Windows's mixer. From past experience, I know that this can make a substantial difference.

I have then played the same FLAC files through Foobar and Roon. It is not a streaming vs local files comparison, but to do that I need to ensure that I can find the exact same version of the exact same album on the streaming platform. There are often a few different versions/remasters of seemingly the same album/song and I think that's the main reason why people hear differences between CD and streaming. Another reason would be that from my experience Tidal' and Qobuz's built-in players sound different than Roon, so either Roon is not bit-perfect or they aren't (I believe that Roon has previously been shown to be bit-perfect).

Well, the same FLAC files played through Roon and Foobar sound more or less the same to my ears using the highly-revealing Focal Utopias. I have not done much listening, so perhaps my opinion will change later on, but at first listen I am not detecting any major differences. So for now I'm happy to say that they're both equally good and I cannot see any reason why streamed FLAC files would sound worse than ones stored on a local disk.

Now for PGGB... I have never used it before. I have downloaded it as a plug-in for Foobar which allowed me to use the software to convert files as I go, instead of doing it offline (I'd expect the result to be the same). I have set the number of taps to maximum 1 billion (what a number ). Immediately I've noticed that the number varies from song to song - it can be, say, 120 million for one song and 149 million for another. Why is that?

Regardless, I immediately noticed that the sound was different. Perhaps a little "fuller" than the filters that I had been using. My experience tells me that there might be many reasons for that so I started experimenting. I switched HQPlayer back to PCM and (bear in mind - I have done very little listening so far) my initial impressions were that PGGB sounded a bit better regardless of whether I used the 2M-tap Sinc-L or my beloved Gauss-long. So far these results have been quite constant so it might be the case that PGGB handles PCM a bit better.

Then I switched over to my preferred setting of the HQPlayer to see if I'd still prefer PGGB - PCM to DSD256 with Gauss-long filter. Initially, it was song-dependent. They sounded different, PGGB reminded me more of the Sinc filters in HQPlayer where sounds were seemingly more full-bodied. And this, I believe, was confirmed when I started listening to tracks where I know there is a lot of depth and where instrument separation can really vary depending on what gear/software I'm using. And then the HQPlayer's DSD with the Gauss filter started to clearly pull ahead, giving me extra depth and presenting each sound as if it had its own space within the soundstage.

I need to do much more listening to reach a result that I could be reasonably certain of; however, my gut feeling and the experience gained so far tell me that what is going on is the following:

• PGGB probably uses different filters which will of course sound slightly different. Some people will prefer HQPlayer's, others PGGB's, since this hobby is very subjective.
• HQPlayer's PCM to DSD conversion seems to benefit the kind of DAC that I have (Delta-Sigma). If I had an R2R NOS DAC, then perhaps things would be different and the apparent (to be confirmed in longer listening sessions) advantage that the PGGB has in the PCM domain would then become an important factor for me.
• The number of taps beyond a certain point does not seem to matter too much. I don't know where that point is but judging that we're jumping from just a couple/few hundred from the DAC itself, to 1 million in Sinc-M, to 2 Million in Sinc-L and then over 100 million in PGGB's filters - I feel quite confident saying that the type of filter (sinc/gauss) matters much more than the number of taps. It's a nice selling point, previously exploited by Rob Watts, but it doesn't seem to matter too much.
• Using FLAC files stored locally, Roon and Foobar sound the same to me. I'd think that the same files (not MQA or differently-mastered versions) streamed through Roon should sound the same as I think it's been shown before that they are bit-perfect FLAC files (as opposed to when, for instance, played via Tidal which has been shown to actually alter the contents of the files).

So, for now it's still DSD via HQPlayer for me.

Just for transparency - I have no technical knowledge to back up any of the above claims. They're simply my personal experience and knowledge gained so far on this journey.

 

[kennyb123]

The number of taps beyond a certain point does not seem to matter too much. I don't know where that point is but judging that we're jumping from just a couple/few hundred from the DAC itself, to 1 million in Sinc-M, to 2 Million in Sinc-L and then over 100 million in PGGB's filters - I feel quite confident saying that the type of filter (sinc/gauss) matters much more than the number of taps. It's a nice selling point, previously exploited by Rob Watts, but it doesn't seem to matter too much.

Thanks for sharing your impressions. I just wanted to comment on the above. The count of taps is most certainly more than just a selling point. Mathematically-speaking, greater taps mean higher reconstruction accuracy. When done right, listeners report sound quality gains. If gains aren’t heard, then it’s either not being done right, something is getting in the way of it being appreciated, or the kinds of improvements delivered by improved reconstruction accuracy aren’t the kind a listener prefers.

Having said all that, I must admit that I haven’t ever heard much of a difference between sinc-M and sinc-L. My take on these is that they were tossed in just to check a few boxes. Rob Watt’s WTA filter via the M-Scaler sounds a lot better, but of course it costs a lot more and one can only get the best from it when using Chord DACs. I do appreciate that HQPlayer has these filters though as they allowed me to be able to sell my M-Scaler and apply the proceeds to a better sever (which matters far more).

It does sound as if your preferred filter is giving you a benefit from noise reduction. A big part of what these filters aim to do is shift noise out of the audio band. Jussi’s filters are excellent at doing that. I believe this contributes to why you hear greater depth and more space around the instruments. It’s likely that the noise profile from that filter better compliments your gear and ears. It’s great that you found a synergistic match. I do think this will vary from listener to listener.

That there is greater separation between instruments is a welcome improvement. That fullness improved is nice too. But what really matters is that what one is hearing has brought them one step closer to hearing a live performance taking place in front of them. Hearing the MScaler for the first time shocked me because it was so transformational (this was even more true with each iteration of PGGB). Brass instruments gained more of the bite that one hears when being in the same room with them. Piano sounds more like the real thing because the hammers hit the strings in a way that’s far more convincingly real. The whack of a rim shot cuts through the air in a way that has one actually looking in that direction to see how the drummer is causing that (I literally did that two nights ago). I could never find a setting in HQPlayer that caused a transformation like this. And when I read the impressions of the fans of HQPlayer, rarely did I ever hear them speaking of the music this way. I think that’s why my first hearing of MScaler shocked me so much. I think 256 bit precision in PGGB might have shocked me even more as each of the qualities I mentioned improved even more. I found myself amazed even how naturally portrayed piano was even with some lesser recording. There’s just a rightness to it that brings me more of the sense I am in the room with it. I realize that it can take a not-insignificant investment in dollars to be able to hear all this. But the good news is that the X26pro is capable of letting it all come through.

 

[Mista Lova Lova]

Thanks for sharing your impressions. I just wanted to comment on the above. The count of taps is most certainly more than just a selling point. Mathematically-speaking, greater taps mean higher reconstruction accuracy. When done right, listeners report sound quality gains. If gains aren’t heard, then it’s either not being done right, something is getting in the way of it being appreciated, or the kinds of improvements delivered by improved reconstruction accuracy aren’t the kind a listener prefers.

Having said all that, I must admit that I haven’t ever heard much of a difference between sinc-M and sinc-L. My take on these is that they were tossed in just to check a few boxes. Rob Watt’s WTA filter via the M-Scaler sounds a lot better, but of course it costs a lot more and one can only get the best from it when using Chord DACs. I do appreciate that HQPlayer has these filters though as they allowed me to be able to sell my M-Scaler and apply the proceeds to a better sever (which matters far more).

It does sound as if your preferred filter is giving you a benefit from noise reduction. A big part of what these filters aim to do is shift noise out of the audio band. Jussi’s filters are excellent at doing that. I believe this contributes to why you hear greater depth and more space around the instruments. It’s likely that the noise profile from that filter better compliments your gear and ears. It’s great that you found a synergistic match. I do think this will vary from listener to listener.

That there is greater separation between instruments is a welcome improvement. That fullness improved is nice too. But what really matters is that what one is hearing has brought them one step closer to hearing a live performance taking place in front of them. Hearing the MScaler for the first time shocked me because it was so transformational (this was even more true with each iteration of PGGB). Brass instruments gained more of the bite that one hears when being in the same room with them. Piano sounds more like the real thing because the hammers hit the strings in a way that’s far more convincingly real. The whack of a rim shot cuts through the air in a way that has one actually looking in that direction to see how the drummer is causing that (I literally did that two nights ago). I could never find a setting in HQPlayer that caused a transformation like this. And when I read the impressions of the fans of HQPlayer, rarely did I ever hear them speaking of the music this way. I think that’s why my first hearing of MScaler shocked me so much. I think 256 bit precision in PGGB might have shocked me even more as each of the qualities I mentioned improved even more. I found myself amazed even how naturally portrayed piano was even with some lesser recording. There’s just a rightness to it that brings me more of the sense I am in the room with it. I realize that it can take a not-insignificant investment in dollars to be able to hear all this. But the good news is that the X26pro is capable of letting it all come through.

Thanks for sharing. At this very moment I'm listening to The Rolling Stones - "Its' all over now" and man... There's no competition. HQPlayer's DSD with the Gauss filter has soooo much more depth than however many taps I set PGGB to. As soon as I switch over to PGGB, everything shrinks and sounds get much closer to each other.

But the same thing happens when I switch to one of the Sinc filters in HQPlayer, so this is not a dig at PGGB - like I said, it's more to do with what filter is being used than how many taps it's got.

I hear clear differences between the Sinc-M and Sinc-L filters to the point that I initially shared my being uncertain about the HQPlayer because the Sinc-M filter added "meat" but wasn't too airy-sounding. At the time that was the only filter I'd tried as everyone seemed to be recommending Sinc-M as one that resembles the MScaler. As for the difference between these two, another factor to take into account is that M is apodising whilst L is not. And looking at the "error counter" displayed on the HQPlayer interface, so many recordings benefit from an apodising filter.

If the new Gauss-Long filter didn't exist, I'd say that both PGGB and HQPlayer have their strengths, although in the case of the Gus I do believe it benefits from being fed DSD signal. But since this filter does exist, there is no real competition for me, using the gear that I've got. Even my dad said: "It's like adding a few metres of soundstage" (about the HQPlayer).

I believe it's got to do with striking the right compromise between the frequency response and time-domain. From what I know, brick-wall filters achieve the best frequency response (those Sinc filters) but they cannot compete with something like the Gauss when it comes to time-domain and it's so immediately obvious that since I discovered that filter I've not been able to use any of the Sinc ones as they all sound "shallow" and congested to me, albeit this might come across as more full-bodied (but that's in the same sense that, for example, the Topping D90 comes across as punchy because it sounds rather 2D and all sounds hit you at once).

The Gauss filter seems to have it all to my ears without any obvious sacrifices/trade-offs. I already loved it in PCM but it's really blown me away in DSD.

 

[kennyb123]

Thanks for sharing. At this very moment I'm listening to The Rolling Stones - "Its' all over now" and man... There's no competition. HQPlayer's DSD with the Gauss filter has soooo much more depth than however many taps I set PGGB to. As soon as I switch over to PGGB, everything shrinks and sounds get much closer to each other.

Thanks for the additional details. Depth and the space between instruments isn’t something I ever pursue. As I mentioned before, it’s all about realism to me. Image size can only be assessed by how realistically the instrument is portrayed. DSD tends to round things out by softening the edges. This has always struck me as taking a step away from realism when it comes to instruments that are struck or plucked.

I don’t think what you say about time domain performance is accurate with respect to approach that either Rob Watts or PGGB aim for with their respective approaches, as both aim for Whittaker-Shannon. As Rob has pointed out in the slide that follows, this filter can perfectly reconstruct the analog signal with zero timing errors. I know nothing about the gauss filters, but I just wonder how they could improve upon what Whittaker-Shannon promises.

I hope the day eventually comes when I hear a fan of HQPlayer speaking of its benefits in terms of the music instead of in terms of the sound (that was meant tongue in cheek). I think though the difference just reflects the different aims some have. We each hear differently and have different musical tastes, so one size will never fit all. It’s wonderful that we have options.

The following is from Rob Watts slides on the MScaler.

 

[EMINENT]

I tried the 256 version last week, but the first run I did was not favorable to PGGB vs. HQP PCM-DSD1024 I am used to.
There are a lot of factors at play here though from PCM vs. DSD and the filter/modulator someone would choose using. I've found the many combinations to have different presentations and the ability to choose between them is the advantage to me as whatever is happening on the PGGB file is stuck as it's rendered. This could be also sounding like a filter/mod presentation that I might not prefer. It definitely has it's upsides for when running a file and including eq or convo on a dap that would otherwise not be able to run real time HQP processing.
Like I said, i'll try once more with the latest update here soon and report next week.

 

[taipan254]

@Mista Lova Lova I'm very happy to hear you have found your "flavor" of upsampling. I'm also glad you gave PGGB a try. There is little doubt that the X26-Pro is a very capable DAC, especially relative to its price.

While I think we come to different conclusions with respect to upsampling (I favor PCM-based upsampling, even on HQPlayer... with the caveat that I do not have enough horsepower to do intense 512x DSD), I actually love hearing diverging opinions because those opinions force me to reexamine my own beliefs. I firmly believe there is no OVERTLY right or wrong answer - everyone hears differently and has different systems (with different power and networking quality).

A few points of clarification:

Immediately I've noticed that the number varies from song to song - it can be, say, 120 million for one song and 149 million for another. Why is that?

My understanding: Each song, in its base format, has a finite number of samples. As a result, there are a finite number of taps that can be applied to each song. More taps are possible when the base material has more samples available to interpolate (more samples happen when the base material is longer and / or utilizes a higher sampling rate like 192k).

Roon and Foobar sound the same to me

This is contrary to my experience, and the experience of others that I trust. I'd be curious to know your set up - are you running your DAC straight from your computer? I started noticing many more differences when I started using an endpoint (ultraRendu) powered by a linear power supply connected via a netgear switch (do NOT use your wifi point's built in switch.... SO. MUCH. NOISE!). This is absolutely not the height of hi-fi set ups, but I will tell you it makes a difference. If you are not running an end-point based system and have an extra computer (even raspberry pi) laying around, I'd recommend setting up that extra computer with free debian and jussi's NAA image, and connect it via a netgear or d-link switch (or bridge this end point to your main PC if your main PC has two ethernet ports). My hope is that your sonic bliss - no matter the filter - will be elevated (and not to mention you'll hopefully hear some differences in software - your focal Utopia and DAC deserve it!).

It's a nice selling point, previously exploited by Rob Watts, but it doesn't seem to matter too much.

It's funny @kennyb123 put up that sinc function slide because I wanted to share something similar as well when I first read the above comment. The theory with the sinc filter / whittaker shannon is fascinating. See attached for a longer discussion. The TL/DR as I understand it: Whittaker / Shannon created a perfect way to digitize an analog signal of limited bandwidth. However, to reconstruct that signal, a sinc function of infinite tap lengths is required. This does not exist in the real world. Many DACs / DAC Chips use limited numbers of taps and noise shaping. This is an economical approach. While this approach gets very close to the original signal, it is not identical to the original analog signal - i.e., there are differences between the "theoretical perfect analog reconstruction" vs. what was actually produced. Most notably, this impacts transient timing. We previously couldn't use long tap lengths because of limited DAC horsepower, and limited computing power for that matter. This has changed. Rob Watts exploited the power of FPGAs to maximize tap length for real-time processing. PGGB has exploited the innate power of our computers by "pre-processing" files / applying sinc filters of crazy tap lengths with very high precision (256 bit math). The difference between the "theoretical perfect analog reconstruction" and what PGGB produces should be far smaller relative to the same difference without PGGB processing.

From a theory perspective, I whole-heartedly buy into this approach. It certainly isn't the only approach. PCM to DSD conversion is another approach (PS Audio, HQP). R2R is another. CH Precision uses splines instead of sinc functions (see Stereophile's recent review of CH Precision's C1.2). There are other approaches as well. I second @kennyb123 's desire to learn more about the filters Jussi uses in HQP (apart from the basic description in the HQP manual).

In any case - that was far too long for a TL / DR! If this hobby has taught me anything it is that some people will pay crazy $ to get close to musical nirvana, but musical nirvana is a "moving goalpost". The craziest (and smartest) audiophiles I've seen on the net are those that bought the Taiko Extreme. When folks first got the extreme, they were smitten with it - to the point where you would think it was perfect and greatly better than any other thing on the market. But then software improvements were made and more improvements were had and folks were MORE smitten with it. Now there are bespoke DAC USB drivers, new switches, routers, USB cards, interfaces, and even a Taiko DAC and those Taiko fans are continuously getting even more enthusiastic (and this greater enthusiasm comes as they pay more and more money for more and more Taiko accessories). I bet you they are hearing amazing things.

I'd love to find a more affordable way to get there though!

 

[taipan254]

I tried the 256 version last week, but the first run I did was not favorable to PGGB vs. HQP PCM-DSD1024 I am used to.
There are a lot of factors at play here though from PCM vs. DSD and the filter/modulator someone would choose using. I've found the many combinations to have different presentations and the ability to choose between them is the advantage to me as whatever is happening on the PGGB file is stuck as it's rendered. This could be also sounding like a filter/mod presentation that I might not prefer. It definitely has it's upsides for when running a file and including eq or convo on a dap that would otherwise not be able to run real time HQP processing.
Like I said, i'll try once more with the latest update here soon and report next week.

Very cool - you're using the Holo May KTE, right? Were you PGGBing to 1.536MHz? What filters do you favor for HQP to DSD1024?

I'm really curious about that DAC, but given financial constraints, I don't see myself buying one any time soon.

 

[EMINENT]

Very cool - you're using the Holo May KTE, right? Were you PGGBing to 1.536MHz? What filters do you favor for HQP to DSD1024?

I'm really curious about that DAC, but given financial constraints, I don't see myself buying one any time soon.

Yes, on the May with PGGB to 1.535. I had X26P before and am quite familiar with it doing 768Mhz PCM and used sinc-Mx with LSS15 back then.

Now it is this.

 

[kennyb123]

While I think we come to different conclusions with respect to upsampling (I favor PCM-based upsampling, even on HQPlayer... with the caveat that I do not have enough horsepower to do intense 512x DSD), I actually love hearing diverging opinions because those opinions force me to reexamine my own beliefs. I firmly believe there is no OVERTLY right or wrong answer - everyone hears differently and has different systems (with different power and networking quality).

I agree about us reaching different conclusions. Having said that, some of these filters are objectively better than others. These have the potential to give us higher reconstruction accuracy - and would help to get us closer to the sound quality that the microphones captured originally. Certainly the ideal Whittaker-Shannon should get us there, but like you said that’s impossible to achieve that in the real world. I do think though that the closer we get to being able to approximate this, the closer we will get to being able to hear what the analog signal sounded like before it entered the A/D converter. It’s less about subjectivity at that point, though of course many of us will still want to season the sound to get it to better align with our desired preferences. There is no right or wrong answer if the goal is our musical enjoyment.

Public service announcement: those testing these filters should check to make sure they’re enabled NOS mode. I inadvertently made that mistake once or twice when piddling around.

 

[J Mirra]

Thank you for bringing Gauss to my attention on HQPlayer.
I hear more depth and layering.

 

[Mista Lova Lova]

Thank you for bringing Gauss to my attention on HQPlayer.
I hear more depth and layering.

I'm glad you're enjoying it as much as I do! It's phenomenal, nothing else comes close on my system. It makes music sound so much more realistic - not by changing the frequency response but by reconstructing more depth which to my ears results in sounds being less congested, less "on top of each other" which allows them to sound more life-like.

And even Rob Watts says that the filter which has more depth will be the more "correct" one (looking it at from the "objective" point of view).

I recommend using the HighRes-lp filter (will double-check the name, not sure if I got it right) for the second stage of oversampling (the selection right below the first one on HQPlayer). This combo is pure gold.