PGGB and HQPlayer Discussion

[Mista Lova Lova]

I've had a wee break from music recently due to personal circumstances and I think that it's done me some good as I'm returning with a fresh pair of ears.

 

First of all - @Zaphod Beeblebrox and @Miska - thank you so much for both of these wonderful pieces of software. Thanks to being able to use different filters, different modulators etc. I have been able to fine-tune my hearing, learn so much about the nuances that would have otherwise gone unnoticed - and enjoy my music in a variety of ways, which is what this is all about. 

 

@GoldenOne's recent post about double-upsampling (PGGB then HQP) has led me to try to convert PGGB-upsampled PCM files to DSD using Sinc-L. I have done some comparisons with HQP PCM and DSD upsampling using the native file. My goal here was personal preference as opposed to chasing numbers (noise shaping etc.) - I wanted to combine what PGGB has to offer with possibly also being able to use my favourite DSD modulator.

 

Basically, my conclusion is that PGGB sounds like Sinc-L on steroids - in a good way. Other filters aside (which sound sometimes very different and where personal preference might come into play), when it comes to Sinc-L type of sound (which I would describe as neutral and spacious, without any detectable colouring), PGGB basically does everything better than Sinc-L. Same/similar idea, just a different level of sophistication. I tend to agree with @GoldenOne that PGGB will bring the most information (and "smoothness") out of good recordings. It has a very "neutral" timbre, doesn't add warmth/colouration like e.g. Sinc-Mx seems to do (it's not just a matter of it being apodising as I detect this shift on tracks with no apodising errors also). Listening to PGGB-upsampled files I certainly get the impression that I have never heard so much information extracted from them.

 

However, I am really struggling to fully enjoy it on my system. I believe it's nothing to do with PGGB but with my DAC and my perception of how it processes PCM vs DSD (sorry @kennyb123, I'm glad that you don't share my opinion here and as such are able to enjoy it without any reservations, but we'll just need to agree to disagree here). Yes, Gustard X26 Pro does not do native DSD and yes - many will disagree with me about DSD sounding better. But despite my wanting to convince myself that PCM is all I need, the HQP 7EC-Super modulator is just a level above. A flawed analogy perhaps, but it's like playing Cyberpunk 2077 with Ray Tracing (especially Path Tracing) off and on, respectively. The former feels like a facsimile of the real thing, the latter really does approach what my eyes (in this case - ears) are willing to accept as real(ish). 

 

Whether this is because, as @Miska says, the ESS's chip is limited in the number of calculations it can do on the fly and as such delivers a less-than-perfect output, or it is simply a matter of taste - I do not know. All I know is how much more convincing it sounds. (The latter is actually quite possible as, for instance, I could not praise the non-super modulator as unequivocally as I am happy to praise the -super one).

 

So, I love what PGGB does in terms of extracting so much information from the recordings (I'd add a caveat that some bad recordings might benefit from one of the HQP's filters and some might prefer the latter altogether for various reasons, such as ease of listening) but I also love what HQP does in terms of helping me deliver the file to my DAC in a form that results in better final output.

 

The problem now is that my initial testing shows that further upsampling a PGGB file to DSD using Sinc-L, whilst still retaining all the information, does not sound so convincing. It appears to suffer from the same phenomenon as whenever I forget to switch the NOS mode on in my DAC, whereby the PGGB- or HQP-upsampled file would then be subject to further filtering using Gustard's internal filters. And that makes things sounds weird - unnaturally smooth, as if there was a layer between me and music, which results in a sound that is less direct, less incisive, kind of "sterile". 

 

The above experience leads me to the conclusion that I have not been able to fully experience what PGGB is capable of (and, for that matter, same applies to HQP really as I can't to DSD Direct; however, I am at least clearly still able to experience some benefits of DSD as the difference is staggering with the right modulator). Is there any way that I could convert a PGGB-upsampled file to DSD without experiencing the above oversmoothing/flattening effect? Or would I basically need to get a good quality R2R NOS DAC to let PGGB spread its wings? 

 

In a way, I'm dreading the answer as it'll probably mean that it might be time to upgrade my gear. On the other hand, the thought of being able to experience both PGGB's and HQP's full potential is very exciting.

[Mista Lova Lova]

4 hours ago, Miska said:

Everything... Much more accurate reconstruction in the end.

With all due respect, @Miska, that is highly unlikely to convince anyone on this forum. I presume that your arguments would be: a better time-domain reconstruction (if it's indeed true that longer filters necessarily wreak havoc in the time domain), faster filter response (if this is indeed audible) and higher attenuation (when compared to Sinc-Mx or Sinc-L).

 

I appreciate all the differences between various filters, from gauss-long to Sinc-L to PGGB - each sounds distinctly different to my ears. I still haven't fully grasped why my preference can very much depend on the track I'm listening to - the inquisitive side of me keeps wondering if this is fully subjective (my ears' sensitivity to certain things and the resultant choices made to counteract this and perhaps also the limitations of my gear where I may be compensating for my headphones' deficiencies). But I can't help thinking that there's more going on here.

 

For example, with a track like Dave Brubeck's "Take Five", I would rank the filters as follows (from worst to best):

 

gauss-long --> Sinc-Mx --> Sinc-L --> PGGB.

 

Each step up brings extra clarity and in this particular track nothing occurs that would annoy my ears or make me perceive the sound as unnatural. Also, in this particular track, space (size of soundstage) matters a lot to me and only the very long filters are able to reconstruct this properly.

 

But then, with a popular track like Fun Lovin' Criminals "Scooby Snacks" or "Come Find Yourself", it's now the opposite:

 

PGGB --> Sinc-L --> Sinc-Mx --> gauss-long (if the last one feels too intimate, I'd recommend gauss-xla).

 

It would seem logical to me that one of the below explanations must be correct in relation to the optimal choice of filter:

 

1. The contents of a particular track (genre, how busy the track gets, transients etc.).

2. The quality of mastering and analogue to digital conversion (see below for Archimago's article on ringing in actual music).

3. My personal preference whereby I am compensating for my gear's and/or headphones' deficiencies by adjusting warmth/smoothness and the size of the soundstage by using different filters.

 

Archimago wrote this piece years ago:

 

https://archimago.blogspot.com/2018/01/audiophile-myth-260-detestable-digital.html

 

where he showed that ringing can in fact occur on poor recordings. I'm wondering if this would at least partially explain why certain recordings appear to benefit from (to my ears) shorter filters and the Gaussian treatment (transients?). Or perhaps it's my gear/headphones that emphasise clarity to the extent that on less-than-ideal recordings I feel the need to offset this by using a different filter. I'm doing my best to stay open-minded here.

 

Given that the same filter (gauss-long) that worked so well with popular music clearly puts the brakes on the level of clarity and crispness that I'm able to get out of the "Take Five" recording, it would appear to me that the only logical conclusion here (other than personal preference) would be that filters such as PGGB, Sinc-L and I guess Sinc-Mx also (if we agree that removing apodising errors may be of audible benefit on certain tracks) are the ones that do actually get the most out of recordings (and as such would be the most "correct" way of reconstructing the original signal) with the caveat that they will also expose suboptimal recordings for what they truly are.

 

If this is correct, then both @Zaphod Beeblebrox and @Miska would be right to the extent that PGGB does the best job of extracing the most out of recordings going "by the book" (which assumes that the recordings are of a good quality, with the results being spectacular when that is indeed the case) whereas HQP provides filters which may yield better real-life results with recordings that are inherently flawed, thus perhaps deviating from "by the book" approach, but if the original content is already flawed then finding unconventional ways to try to rectify it may result in a better overall experience.

 

Another flawed analogy from the video world which hopefully helps to illustrate my point - apparently the only "high-fidelity" sharpness setting for your TV is "0" as it stops your TV from further processing the image that it receives from the source. Makes sense - from a technical point of view any interference surely results in some form of quality degradation/artifacts when compared to the native image. However, in practice not all native footage will look best without any sharpness added, especially when viewing, say, 1080p video on a 4K TV. So, it would appear that both things can be true at the same time - further processing technically deviates from high-fidelity as per "by the book" approach, but in some cases yields results that some of us may interpret as more life-like.

 

Just a loose analogy, of course; my attempt to make sense of what I'm hearing by looking for an answer that explains why both approaches yield different results with different source material (to my ears).

[Mista Lova Lova]

12 minutes ago, Miska said:

So you think I don't know technical / objective aspects of filters I've designed?

Not at all. You're taking my comments personally, probably because you're proud of your creation (and rightfully so) and feel the need to defend it. I think you're perhaps struggling to understand things from my perspective - someone who doesn't have first-hand theoretical and practical knowledge about digital audio reproduction. I am therefore not challenging you in terms of claiming that I am right and you are wrong. I am merely trying to get you to share your knowledge more and in the process see how many claims are verifiable facts and how many are "educated guesses".

 

When I see competing ideas where one side claims that a given way of digital audio reproduction gets as close to "perfection" as we have so far and the other side presents a counterargument claiming that the other approach only "perfects" one aspect of sound and ignores all the trade-offs that this creates - I have no way of verifying who is right here as, to my knowledge, neither side has been able to present clear evidence that they're right (e.g. where one side claims that there is "ringing" which smears the time domain, the other will reply that this ringing is irrelevant as it occurs outside of our range of hearing etc.).

 

So the only thing that I can do is listen with my own ears. And, unfortunately, things are complicated here too, because, as referenced above, it is clear to me that the very long filters extract more data (space, holography) out of the digital file, but certain recordings sound better using different filters. My inquisitive nature makes me want to keep digging to establish if this is just my personal preference or if digital audio is indeed so complicated that trade-offs are always involved, except perhaps for the very best recordings.

 

25 minutes ago, Miska said:

Just curious, why this specific set out of all the options? Why not for example just gauss, gauss-short, or ext2 for example? Why Mx and not MGa?

 

Good question and a fair point. I could, of course, include more comparisons; I picked the filters which all sound convincing to my ears, despite very obvious differences between them.

 

Why not gauss-short? Very simple answer - it's just rubbish to my ears. Might be good for "snappiness" of transients and for experimenting with filters so as to fine-tune one's hearing by finding out what's missing if we make a filter too short. So absolutely no issues that it exists, it fits perfectly with HQP's philosophy of giving the listener a wide range of filters to choose from. But even going by the manual - this filter only focuses on transients, not timbre or space - and the last two aspects are very lacking indeed. For some specialist usage - yeah, perhaps. For general listening to music and wanting it to sound convincing - no way.

 

Why not the "standard" gauss? Even though it improves timbre compared to the short filter, space is seriously lacking here. I personally don't find any filter shorter than those going up to 20kHz to sound convincing because, to my ears, space is a very important part of music reproduction. Otherwise I just cannot be convinced that I'm listening to real sounds.

 

Why not ext2? Beucase, as per the manual, its only focus is on timbre and that shows, too. Space is lacking and the filter sounds rather "thin", compared not only to gauss-long but also to Sinc-S.

 

Sinc-MG (and MGa) sounds to me like Sinc-Mx but with softer transients. And again my experience has been that this is a double-edged sword - it does help some recordings but sucks the bite/punch out of some others. Sinc-Mx sounds "fuller" at the cost of being sharper, Sinc-MGa sounds a little "thinner" but "takes the edge off" which can be pleasant. 

 

It seems to me that the main issue we're facing is how to take a filter like the Sinc-L or Sinc-Mx and increase its transient reproduction/accuracy/timing further. And here we appear to have two different approaches: either add a Gaussian element to it and preferably shorten the filter (if its length does in fact affect the time domain) or extend the filter's length further and thus potentially increase its precision without adding the Gaussian "softness" or sacrificing the frequency domain in the process.

 

I am not arguing that one approach is correct and the other one is not. Just trying to sum things up here the way I see them.

[Mista Lova Lova]

56 minutes ago, Miska said:

I personally mostly stick to the current default, as I've found it to be most universally applicable:

1x = poly-sinc-gauss-long

Nx = poly-sinc-gauss-hires-lp

 

That's usually my selection, too. However, having returned to the hobby with a relatively fresh pair of ears now, I must admit that this being a "safe"/middle of the road approach, it can make bad recordings sound much more palatable but it also keeps the good ones from spreading their wings. Which, again, makes me wonder what it is about those recordings that is different. Perhaps it's the analogue-to-digital stage that wasn't done properly, or perhaps it's about the content of a given track (are instruments used or are the sounds electronically generated etc.). And if the latter - is the ringing really so prevalent so as to affect the final output so much (again, there are claims that this is a non-issue as it only really occurs with "illegal" signals and near Nyquist anyway).

 

30 minutes ago, Miska said:

sinc-L also has worst time domain performance while poly-sinc-gauss-long has best

 

Do you mean that the gauss-long has the best time-domain performance of all of the HQP filters? Given the relationship between the frequency and time domain performance described by you, should filters such as gauss and gauss-short not exhibit an even better time domain performance (albeit at the cost of further sacrifices in the frequency domain)?

[Mista Lova Lova]

My listening tests continue (which will probably always be the case, as it's a hobby of mine that never seems to go away for too long).

 

I have done all I can to approach this with a fresh perspective and following a recent break from audio I tried to "get my mind used to" PGGB so that it would treat it as a default option (as opposed to having fresh muscle memory of my usual HQP settings). However, no matter how long I listen, I seem to always come to the conclusion that in addition to certain tracks sounding better with an apodising filter (let's put that aside for now), PGGB'd files seem to have transients/edges of attack that sounds too soft for me. I can see how some might find this sounding preferable, perhaps viewing it as sounding "correct", but I could never part with HQPlayer precisely for the reason that those transients literally get a physical reaction from me whilst I'm listening (finger-tapping etc.) which is gone with PGGB which tends to present transients as if they were wrapped in a soft pillow, which at first seems nice and "cosy", but in the long run it's like watching an HDR movie with the HDR mode switched off. Maybe it's easier on the eyes in certain situations but you're missing out on so much nuance (talking about transients only, not space etc. which PGGB gets right).

 

For my fellow PGGB listeners - to get what I'm talking about, try using HQP with Sinc-MG filter (not MGa, so as not to introduce apodising as yet another variable). Shorter filters will add variables which will make comparisons even more difficult, hence I believe that Sinc-MG may be the perfect candidate for comparisons. This filter does reproduce the entire frequency range and to my ears roughly the same perception of "space"; but transients sound very, very different. I was just listening to Dr Dre's "2001" album and couldn't get enough of those "edges of attack".

 

To use another (poor) video analogy: my in-DAC filters are like pre-anti-aliasing image - quite rough. PGGB (sometimes) sounds like anti-aliasing that's been overdone, as if some "blurring" was introduced which gets rid of the roughness of the original image but does it at the cost of taking some of the incisiveness away. HQP's Gaussian filters are somewhere in between. I tend to agree with @Miska that even shorter filters (such as gauss-long) get the "snappiness" of transients even more right. But then it's a matter of trade-offs between other aspects of sound. To my ears, even Sinc-MG sounds significantly different from PGGB when it comes to transient reproduction.

 

Another thought I've had recently, and it involves another video analogy (sorry, been playing too many games recently): I think that I've found a way to convey what I like (and don't like) about Sinc-L and Sinc-Mx. They seem to be the best at getting the "body" of each sound right - what I would refer to as "full-bodied sound". However, and perhaps this ties in with some of @Miska's comments, this only really works well when not too much happens in a track. When there's a lot of sources of sound, when they interact with each other within the soundstage and when there's a lot of transients involved, they suddenly stop sounding so convincing and their lack of "flexibility"/"plasticity" becomes apparent. To continue my flawed video analogy - a great-quality image is produced as long as it's relatively static. When we introduce lots of movement, the bubble bursts to some degree as the "suspension of disbelief" (the effectiveness with which my brain is convinced that it's watching/listening to an actual event) is greatly diminished.

And this is where for me the Gaussian filters come in which are like a sauce that make a dish more than the sum of its parts. They don't quite reach the full-bodiedness level of the pure Sinc filters, and perhaps that's been the reason for my frustration when looking for a "perfect" solution.  When it comes to this aspect of sound, it would be great to have the cake and eat it, too. But, despite no solution being perfect, for the music that I listen to it's the Gaussian filters that usually provide a much better listening experience.

 

One last thing that I'd like to suggest to anyone who's reading this and doing their own comparisons - please try the poly-sinc-gauss-hires-lp filter when listening to hi-res content (88.2kHz and above, such as Daft Punk's "Random Access Memories" album), and if you have a Delta-Sigma DAC I highly recommend that you use DSD mode with the EC7-Super filter.

I don't think that I've found a solution to be 100% perfect satisfied when playing a Red Book file (although playing it twice using a shorter and longer filter can be fun, too), but when it comes to hi-res stuff (more frequency range to play with when designing a filter) I have simply not heard anything better than the above setting. It makes things sound both big/spacious and natural/effortless at the same time; it gets both the macro and the micro right at the same time. It's the closest to my idea of "perfect" and I'm wondering how it's perceived by fans of PGGB.

[Mista Lova Lova]

1 hour ago, Miska said:

Why this and not one of the recommended ones? For example ASDM7ECv2, ASDM7EC-light/super, or something like that?

 

My two cents: choosing the right modulator makes a very significant difference in my experience. That's why in my post I specifically referred to the 7EC-super modulator (I wrongly called it "filter" - I wasn't able to edit my post afterwards) as for me that's the one that sounds best overall. I personally wasn't a fan of the 512+ modulator, so as per @Miska's comments, I would strongly suggest to give -Super a try, if one's DAC reacts as well to DSD as mine does (to my ears). It not only sounds bigger and more impactful in the macro sense, but also removes "grain" (for lack of a better word) that I sometimes perceived some of the other modualators (e.g. ECv3 to have). It's smoother but not overly so.

 

@austinpop I can really relate to your quote from the aforementioned article:

"Technicalities aside, the upsampled DSD256 tracks supplied another large step up in sound quality. The sound was more lustrous and refined, with excellent bass heft. There was a real sense of space and ease, as if the music had been freed and allowed to breathe. Another word that kept coming to mind is natural. Just lovely."

 

That's how I generally feel about my DSD experience (and my DAC does not even support DSD Direct). However, I note that you found PGGB-upsampled PCM to surpass that. It would be interesting to see if spending more time with the set-up would have changed your mind - just like I am always easily impressed by very-long filters (and can also be initially impressed with PCM's seeming punch/wall-of-sound effect that I get on my DAC), but end up using others for my daily listening as the former don't stand the test of time.

 

In any case, if I'm not mistaken, the aforementioned DAC would seem to actually prefer DSD if it's got a dedicated 1-bit conversion circuit and uses Burr-Brown chips for Delta-Sigma conversion (PCM). And yet you still preferred PGGB. I don't think that I've quite been able to enjoy the PCM results on my system to that extent, past the initial impressions which again didn't stand the test of time.

[Mista Lova Lova]

2 hours ago, Schafheide said:

@Mista Lova LovaWhat an absolute intellectual pleasure it is, for me to read your thoughts.  We need more like you, to question our opinions on the many, many variants in this, our chosen love of recorded music.

 

Thank you very much for your kind words. English isn't my mother tongue so I can only hope that I have been successful at conveying my thoughts to others, and articulating those can be quite tricky in this situation where I have even resorted to analogies/metaphors to try to paint a picture of what I had in mind. 

 

I'm glad that what I'm doing hasn't been misunderstood as trying to convince others that my ears are better than theirs and/or that my way of listening to music is "more correct" than theirs. It's just my inquisitive nature that kicks in when I see strengths, weaknesses and the associated trade-offs and I then feel the need to at least endeavour to understand what the underlying reasons for this are. If it was something that I didn't personally care about, I would have perhaps been more easily satisfied with an answer that "kind of makes sense"; however, because I absolutely love music and the positive impact that it has always had on my life (and I seem to be quite sensitive to various aspects of audio reproduction), what would have otherwise been a bit of a "chore" has suddenly turned into a fascinating journey.

 

Having said that - however fascinating the journey may be, at the end of the day it's all about music enjoyment which is inherently subjective. Experimenting is great and I'll certainly continue to do so; however, overthinking surely leads to the "can't see the wood for the trees" syndrome.

[Mista Lova Lova]

On 1/3/2024 at 11:03 PM, copy_of_a said:

„There’s no one-filter-fits-all.“
Poly-Sinc-Gauss! 😊
It really works fine for everything.
A joy to listen to with all kinds of music (through loudspeakers). I love it! It‘s so damn clean!

I don't usually use that filter, I find the gauss-long more convincing overall (some extra space), but:

 

1. When listening for particular aspects of sound in "isolation" (e.g. transients), I find the -gauss more convincing (in relation to said aspects), so this would yet again seem to prove the correctness of @Miska's observations with regard to the length of the filters and the associated properties thereof (this doesn't change the fact that one person may prefer different approach/trade-offs to another).

 

2. There are tracks that clearly seem to immediately benefit from shorter filters (talking about Red Book).

Check this one out: "My Red Hot Car" by Squarepusher. The transients' snappiness/incisiveness with the -gauss and further with the gauss-short (which I would normally never use) filters is simply phenomenal. I would normally still listen to this track with the -gauss-long filter, but then I'll often re-play it with either of the aforementioned two just to soak up those transients.

 

PS I've not listened to the Porcupine Tree tracks yet, but will do so soon.

[Mista Lova Lova]

1 hour ago, semente said:

The problem with your view is that it is subjective. Unless you are able to statistically prove that most (e.g. 90%) HQPlayer users agree with your opinion.

 

I'm afraid that that's the "problem" (or inherent characteristic of) with every view and you are thus expecting of me what is not only unreasonable but also undoable. I would no doubt be a rich man if I were able to produce "objective views", especially on such person-dependent matters as one's perception and enjoyment of a given digital audio filter.

 

If there was ever any doubt as to my reasons for sharing my highly subjective notes on individual filters, it should have been dispelled by the last sentence of the post quoted by you:

  

On 12/21/2023 at 4:31 PM, Mista Lova Lova said:

I am not arguing that one approach is correct and the other one is not. Just trying to sum things up here the way I see them.

 

-----

 

1 hour ago, semente said:

And there's also a chance that your expectations and assessment have been influenced by what you have read.

 

Indeed, I would say that this chance is quite high, hence my attempts to give myself a break from music and approach things with a fresher perspective. Does that mean that I'm suddenly 100% objective? Of course not. But that's the best that I can do and I think that it's still worth something. One thing I know for sure - my love for music/audio is vastly greater than whatever affinity for a given product and the corresponding philosophy/approach I may have (whether hardware of software). So I'll happily abandon the current one as soon as I've found something better- as has been the case throughout my entire audio journey, during which - especially at the beginning - I found my vocabulary lacking in terms of finding ways to illustrate the differences that I was hearing. My hope is that my posts may assist some in doing exactly that - and those views can then be exchanged and either some consensus reached or at least an appreciation of our differences acknowledged.

 

Note that the above is only possible when we acknowledge and accept that all of our views on what we are hearing are inherently subjective. Attacking one's views on the grounds that they're subjective is pointless.

 

Having said that, there are two creators of two great pieces of software on this forum (HQP and PGGB) who, to the best of my understanding, appear to disagree mainly on the topic of the best way to reconstruct the time domain and transients (among other things, such as noise-shaping etc.). I see this as a fundamental schism, so to speak, in the world of digital audio reconstruction and I'm intrigued by it, especially since it would appear to be that both parties cannot be right at the same time.

Do note, however, that I do not have sufficient knowledge and expertise to offer any arguments and/or pieces of evidence which would support one or the other. The best that I can do is lots of critical listening and then try to draw my own conclusions accordingly.

[Mista Lova Lova]

Since it was me who started this conversation, I feel like it's appropriate to give a wee update (or a wee teaser really, at this point) in light of the upcoming release of the DSD version of PGGB.

 

I will refrain from posting detailed impressions until the product is officially available; however I can say this for now - I have never said "oh wow" so many times whilst listening to music before and the last time I got so emotional with how beautiful music had become was probably when I upgraded from Topping D90 to the Gustard X26 Pro DAC. For the first time since starting this hobby, I have stopped using the word "trade-offs" when thinking about the best description for what I'm hearing. I am literally speechless every time I play a gargle-blasted DSD file on my system and it's like everything I've ever wanted and then cranked up to eleven. Digital audio has never sounded so good before.

 

I am so excited that this product will be widely available very soon. The level of one's appreciation for what it does will probably depend on what gear they're using (I know that my Focal Utopia and Hifiman HE1000se are high-quality headphones, but I'm sure that there's more to be had), however in terms of comparisons vs everything else that I've tried to this point, the verdict is clear - the king has arrived and his throne is at least a level above the competition.

[Mista Lova Lova]

35 minutes ago, LowOrbit said:

MLL

 

I liked your post (be rude not to) but you're only fueling my impatience to get my hands on the finished release of PGGBDSD. I heard a few test tracks but without the EQ which would have made evaluation quite a bit more useful to Zaphod and to me. 

 

(Good to know there's more than one PGGB fan in the UK too!)

 

I'm quite convinced that the wait will be worth it. Pay attention to the spatial nuances in terms of depth and what happens at the "outskirts" of the soundstage where there's so much more air around particular notes as if a camera lens's zoom has been upgraded to capture the entire picture with the same sharpness. Spectacular.

 

I personally don't use EQ so don't have any thoughts to share on this.

[Mista Lova Lova]

23 minutes ago, Miska said:

Not related to HQPlayer though... So maybe the PGGB thread would be more appropriate place to talk about that?

 

I think that it is very related to this particular thread which was separated from the PGGB thread in the past for this specific reason. I was a long-term HQP fan and I think that it would be interesting to see what other music lovers' impressions will be.

 

I will of course be guided by this forum's moderators as to what discussions are appropriate but personally I can't see anything wrong in exchanging opinions. After all, I used to praise HQP even when compared to PGGB in the past and I don't recall anyone having issues with those posts (except for members who disagreed with my views, but that was just part of the normal discussion, i.e. exchange of views and opinions). I believe that it's only fair to continue to apply the same criteria here.

[Mista Lova Lova]

You're ascribing bad intentions to what I'm doing, @Miska, and I don't like that.

 

I'm not "targeting" anyone; I'm merely inviting everyone to have a discussion as I'm curious if my views will be shared by others, and if not - why not.

 

This is what I've been doing from day one, I don't think that you had a problem with this as long as I was stating my preference for HQP or as long as I was still making references to HQP still offering something that PGGB didn't. And it still does - on-the-fly processing, ideal for streaming. But in terms of audio quality - to my ears, that's no longer the case. Just like I was spreading my excitement for HQP in the past, I'm doing the same for PGGB DSD now - not because of my affiliation with the former or latter, but because of how it's elevated my music experience. You may not like that - and that's perfectly understandable. But by posting such comments you're only going to attract more attention to this discussion. So, in a way - thank you, I guess.

 

[Mista Lova Lova]

I understand your arguments, however my counter to that would be that the reason why the discussion has always been worth having is that:

 

1. These two products are both software products as opposed to hardware offerings and are widely considered to be the best options available to audiophiles looking for a software upsampling solution.

 

2. As per the numerous previous discussions, there have been serious disagreements between the designers of both products where it appeared to be quite clear that the two approaches were so different that one could see them as mutually exclusive, i.e. either one or the other is right and this should be possible to determine via listening sessions. And this is exactly where the discussion about subjective impressions is the only way to gauge what the general perception of the music lovers such as myself is. If the engineering assumptions behind PGGB are wrong, the sound quality should not be improved, or - more likely - should be degraded. If only some of the assumptions are right - then certain aspects of sound (such as soundstage) should see an improvement and others (such as transients) should suffer in the process.

 

For the above reasons, I would be of the opinion that this discussion should be allowed to be continued. However, let's of course leave it for the moderators to decide.

[Mista Lova Lova]

6 minutes ago, The Computer Audiophile said:

If I’m not mistaken, this thread was initially split off from an HQP only thread or from a PGGB only thread, in order to facilitate a discussion about both products. 

I believe that it was the PGGB only thread where I was comparing PGGB to HQP, doing my best to describe the differences that I was hearing and the strengths that I saw in both products. That led to a new topic being created.

 

6 minutes ago, The Computer Audiophile said:

A further discussion about PGGB DSD is ok, if it discusses it in relation to HQP or in that neighborhood. If solely a discussion about PGGB DSD, then a new thread about that topic is appropriate. 

My intention is to do some comparisons between PGGB DSD and each of the HQP filters which I previously used to describe my impressions and hopefully compare it with other people's.

[Mista Lova Lova]

Roger. I'll make sure that I only post comments in here which relate to both products at the same time. I will prepare a longer post gathering all my thoughts and impressions, to keep things neat and tidy and post it once the new version of PGGB has officially been released so that everyone else can make the same comparisons.

[Mista Lova Lova]

A very interesting exchange of arguments (and some sarcastic comments).

 

The most interesting part to me was @Miska's claim that his focus is on restoring the signal the way it was before it entered an ADC, thus to my understanding criticising PGGB's approach as being inherently flawed in that no matter how close to perfection its reconstruction may be, it will inevitably contain and amplify the errors contained within the ADC-processed signal (if my understanding of what's been said is correct).

 

Prior to the release of PGGB DSD I found the above explanation convincing as the Gaussian filters did to my ears offer a more natural/smooth sound than the other HQP filters (albeit not as full-bodied as say Sinc-Mx). I used to always refer to my choice of filter as "my preferred trade-off", which implied a slight disappointment that I couldn't have it all at once. This has changed now as PGGB has better transparency than Sinc-L, similar or better tactility than Sinc-Mx and it manages to be very smooth, as in natural, without needing to give up any of the full-bodiedness in the process, like the Gaussian filters would do. And all of this happens at the same time which I've never encountered before.

 

Which makes me wonder if PGGB's new reconstruction/upsampling approach simply escapes the previous definitions and allows for minimal trade-offs to be made at the cost of a huge amount of processing being required per track.

 

It's also occurred to me that both @Miska and @Zaphod Beeblebrox could perhaps be right at the same time, to some extent anyway, in that there are perhaps inherent flaws in any signal that's been put through an ADC but not as much of it is audible/relevant as @Miska's approach would suggest. Perhaps a ringing analogy would be adequate here - if in fact ringing is mostly a non-issue, is it worth creating shorter filters which reconstruct a smaller portion of the frequency spectrum in order to prevent potential ringing where in reality, on average, we may be throwing away more than we're gaining? Same with apodising errors - perhaps they are audible in an irritating way in certain songs, but there's also a loss of transparency involved when switching from Sinc-MG to Sinc-MG(a). Are we doing more good or more damage in this process?

 

It seems to be hard to find an answer here that would convince everyone, so the topic is definitely complex. But it's my personal opinion that circumstances have now changed as PGGB DSD's sound quality is so outstanding that it can defend itself against otherwise purely theoretical claims.

 

Which gives @Zaphod Beeblebrox a very strong argument; not a theoretical one but a very practical one - the results are simply fantastic. This may change the conversation from "What is likely to give us a better result" to "What are the actual limits of digital audio reconstruction and how far can we push it".

[Mista Lova Lova]

5 minutes ago, StreamFidelity said:

6 hours ago, Miska said:

And doing so anonymously, behind a pseudonym and seemingly without established company behind his product.

A good point. The personal details are available on Miskas website Signalyst. On remastero not.

 

Come on guys, you're better than that. Ex cathedra and ad personam arguments will not get us anywhere and I'm sure that we're all too old for playing the "stroke the ego" game.

[Mista Lova Lova]

Just now, Zaphod Beeblebrox said:

No, the previous differences you heard was because your DAC mangled the PCM path a bit more than DSD and was unfortunately attributed to unrelated reasons regarding audibility of aliased signals. Nothing has changed in the reconstruction algorithms. I can as easily create a 512fS PCM output and use a 64/128/256 bit noise shaper instead of a single bit modulator, if a DAC can play it. The main change I had do was related to memory handling and paging.

Sorry, I meant PGGB's new approach as compared to all the HQP filters that I used before. I am aware that this approach was already utilised in the PCM version and I was able to appreciate some of its benefits back then, it just wasn't good enough for me overall as I had a clear preference for feeding my DAC with DSD content. Now PGGB has been able to truly spread its wings on my system which has been obvious since the first listen.

But my above reference was in relation to your current approach to upsampling and how it is clearly different from everything else that I've heard, for the reasons stated above - it combines strengths of various different filters that I'd heard before with seemingly no audible trade-offs.

[Mista Lova Lova]

1 minute ago, Zaphod Beeblebrox said:

I don't think you said anything wrong. There are tradeoffs, and preferences play a huge role. Since DSD and PCM do not take exactly the same path, it is really apples to oranges.   It does not tell a whole lot about the quality of the reconstruction algorithms used.

 

But when you switched to DSD, it was a more level playing field.

 

While from your preferences perspective, DSD was always your choice, there are many who preferred PCM over DSD on the same DAC. And similarly, may or may not prefer PGGB on your DAC.

 

In summary, my approach has not changed, but the path through which it is delivered has changed for you.

I was under the impression that PGGB used to initially adopt a different upsamling approach (simply a very, very long filter) which at some point changed, hence your previous insistence that describing it as simply an ultra-long filter wasn't accurate/wasn't doing it justice. I was referring to this in the context of the above exchange of arguments - HQP to my ears offers filters which have particular strengths and as such it is very easy to understand why different people may prefer a different filter. Your product is the first one which has provided me with an experience where I don't feel like I'm sacrificing X for Y - I feel like I'm getting everything at once, cranked up to 11. 

 

And yes, like you say - getting it in DSD was for me personally the last, but also a very needed step towards being able to enjoy it without any ifs or buts.

 

But in the context of the aforementioned exchange, I was expressing my supposition that there's something special about PGGB's approach as it delivers a new level of audio quality (and, as mentioned above, it's not just a matter of the filter being a very long one).

[Mista Lova Lova]

9 minutes ago, Zaphod Beeblebrox said:

I guess it depends on when you last heard PGGB PCM. Previously I did use Windowed sinc based filters that were almost as long as the track. This was with v3, but all that changed beginning of 2023 when I released v5 and the reconstruction algorithms have not changed much since then.

Gotcha. It would have been a while ago so I probably heard the pre-2023 PGGB first. The current version in its DSD form is like nothing else out there, to the best of my knowledge.

[Mista Lova Lova]

16 minutes ago, Miska said:

Yes, short filter keeps the transient's energy concentration as close as possible to the transient without smearing the energy over time.

I'm just a layman but I interpret your words as saying that higher pre-ringing is now desirable as it's not an undesired artifact but part of the actual musical content (the "energy" of the transient) that needs to be preserved (in the form of pre-ringing, I take it) as opposed to minimised.

 

EDIT: Assuming that this is even relevant at all in light of @Zaphod Beeblebrox comment and previous discussions.

[Mista Lova Lova]

Again, I'm working on the assumption that this is actually relevant in the audible range (because, obviously, if this occurs only at or close to Nyquist then it's a purely academic discussion) - would it not make sense to have low-magnitude energy spread over a longer time frame, with this energy likely being masked by the actual musical content, as opposed to having a high-magnitude signal present within a short space of time, thus increasing the risk of it actually being audible?

[Mista Lova Lova]

10 minutes ago, Miska said:

 

Of course minimizing the time spread and maximizing the preserved transient energy near the transient is what is wanted.

 

From time domain point of view, those polynomial interpolators give best result, but they do really poor job in frequency domain. So it is about trying to come up with something as short as possible in time domain while being as effective as possible in frequency domain. IOW, considering much much you can "condense" a filter. Optimal case would be a filter which is a Dirac pulse in time domain, and equivalent of infinitely long in frequency domain. But such is mathematically impossible.

 

Although too abrupt changes in frequency response sound bad too, you can easily hear this with room correction filters, if you make too sharp adjustments to the frequency response it begins to sound really bad. Gentler the adjustment, better it sounds.

 

Thanks for sharing your way of looking at this.

 

I have so far looked at pre-ringing as an artifact which is present due to the band-limited nature of the signal but totally unwanted. It seems to me that you look at it as part of the original impulse, i.e. the presence of pre-ringing necessarily takes away the energy from the actual signal and as such degrades that signal even if it's inaudible as such (correct me if I'm reading too much into this).

 

On paper, it seems to sound logical/plausible. But do have a listen to PGGB 512DSD - I don't think that you'll be able to hear any loss of transient energy. Do make sure to take into account the difference in spatial presentation, i.e. depth. Shorter filters can come across as having more energy because the soundstage is simply flatter and certain sounds appear closer to the listener. But it would be like saying that the Focal Clear headphones have better transient energy than the Focal Utopia because they lack the depth and holography of the latter.

 

And if your theory is still correct, then at least it only seems to apply to very high frequencies which perhaps explain why I've not been able to notice any degradation of transients with PGGB DSD (quite the opposite, actually).

[Mista Lova Lova]

Looking at the latest posts from @Zaphod Beeblebrox and @Miska, I can see clearly now that we are actually starting to go around in circles again and this is definitely not good for anyone and not moving the discussion forward in any way. I've certainly learnt something here, though, and because of that I am personally glad that this little debate did take place.

 

Can I just thank you both for publicly sharing your arguments and counter-arguments here, I'm certain that they will be of some use to the community in the future.

 

If I can please suggest that we now focus on posting our impressions, preferred settings and ideally a wee explanation as to what it is that we find particularly appealing about those. 😃

 

I'll start by posting mine soon.