Final Thoughts
Contents
Although the Hugo M Scaler is still my preferred upscaler, the HQPlayer’s capabilities are undeniably impressive. If you enjoy a liquid sound, neutral-greyish tone, with analog solidity…you’ve just struck gold. HQPlayer is one of the best “bang for the buck” products in this industry. You’ll get a ton of value from HQPlayer as it is not only affordable – but serves as a transformative complement for any DAC. With its long list of filters – there’s something for everyone.
The common theme with these upscalers is the loss of warmth as you increase the multiplier. Hopefully, we’re able to preserve some of that richness while reaping the benefits of depth and dimensionality in future technologies. Currently, it seems like this is the expected tradeoff.
This is just one man’s opinion. If there’s an HQPlayer setting you’d like me to check out, please let me know below. I’m sure HQPlayer will continually improve and I’ll keep you posted. As always, evaluate with your ears and form your own opinions.
Did you try Sinc-L with LNS15, DSD256 with ASDM7EC and DSD512 with Poly-Sinc-MP ?
I’ll have to load in some DSD files (as mentioned, I rarely listen to them). But after a quick listen on the TT 2 with sync-L and LNS15 (Cannonball’s Spontaneous Combustion), HQPlayer sounds a bit stretched horizontally and vertically (the snaps are further off to the left and the ceiling seems lower). It sounds a bit off. In addition, HMS is far more tonally variant and layers out the performers better. You could actually hear the organics of the finger snap and the finger plucking the bass. There are more reverb and resonance overall to give you an idea of the acoustic space. The dips, curves, and flair of each instrument are clearly in view. But not so much on HQPlayer. It’s more bundled together and “gel-ed.” Everything has the same tone.
Hi Jay. First, thanks for the time and effort here- great review, especially your perceived differences across the wide range of PCM settings. In response to this specific comment, it looks like you have not tried the DSD settings because you rarely used DSD / have DSD native files? If that is the case, you and your readers might be missing out on a great feature of HQP – which is playing PCM files via the DSD settings, which is how I use it, either redbook rips or my own rips from vinyl at 96/24. In all cases, this was better than any PCM setting (and anecdotally, my preferred settings were the same as yours for PCM). The ‘EC’ shapers that were added only recently to HQP are a big step, in my opinion, over any PCM setting, and definitely worth trying. Whether this narrows the gap to M-Scaler (or exceeds it) i dont know, but as a chord DAC user myself, I no longer feel a need to add the M-Scaler to my system, and am looking instead at upgrading my AMD PC to a chip that will manage the DSD settings at a higher bitrate than my current limit of 128k
VERY intriguing. I’ll give it a shot. Which filter are you using? Seems like most people are using sinc-L? If this PC can’t do EC 512…I’d be surprised. lol
I am restricted to the more straightforward filters if I want to use the more CPU-Hungry shapers, and as I think the 7EC is the best for my taste I use poly-sinc-mp for DSD, and Sinc-m for PCM (although I never really use PCM these days). Like you, the linear filters were not so much to my taste. Yes, that’s one powerful machine you have there, so 256 would be a breeze and I hope you get 512 as well – very interested in what you think one you get a chance to try it out (set aside a while day- 2 mins to change the filters and the rest to just listen and enjoy 🙂
Thanks for the detailed review, Jay. Very useful. Comments: (1) IMHO, the biggest gain of HQP is upscaling to DSD256 using EC Modulators on Delta Sigma DACs (I highly recommend the inexpensive RME ADI-2 DAC FS (Version 2) as that is the one that Jussi, the HQP developer, uses. Don’t forget to enable DSD direct on the DAC (2) There have been quite a few posts online of sellers of MScaler / Dave because they found optimally used HQP to be better (3) A $800 M1 Mac Mini is fantastic host for HQP desktop as with that, I can upscale to DSD256 with ASDM7EC modulator with poly-sinc-ext2 filter without any issues at all on my RME ADI-2 DAC FS. Try it – it might save you a whole lot of money in the long run.
And yes: with the M1 Mac Mini, I don’t get the latency you mentioned before playing tracks.
Great write-up. Would love to see this type of thoughtful assessment of HQ player applied to a “regular”
non-Chord DAC. Since Chord uses a highly unique method of DA conversion its not that surprising that a computer specifically optimized for this process (M-scaler) will beat a non-optimized computer.
Agreed. I’ll do more testing once these listening rooms are done.
Fantastic comparative review, and as an HQ Player user I really enjoyed you assessment of the various filters and op[tions, These were incredibly concordant with my impressions as well after months of use, well-done!. I do think you may be a bit unfairly off putting potential HQ Player users with the statement “If I can’t be convinced of what I’m listening to is a piano or saxophone…” as it relates to HQ Player’s articulation of instruments. Surely that is hyperbolic and whatever smoothing of tonal variation the software produces is far more subtle and discriminating then would hinder that distinction.
In any case, thank you for this excellent read!
The sound of a hammer hitting a string on a piano and the “brassy” sound of a trumpet or saxophone just wasn’t there for me. I don’t hear the fiber or the inherent resonances of the bass plucks or drum skin flex. It just sounds like HQPlayer is more heavy-handed. You don’t hear how the sound comes out of the instrument with HQPlayer (IMO) – which leaves me unconvinced. But you could feel the mass and strength of that voice or instrument. That’s what I meant.
The value proposition laid out in detail that so many of us have been waiting for…
Very nicely done.
You’re connecting the PC running Roon Core (ROCK) directly to your DAC with one of the standard USB ports on the PC?
I’m connecting one of the USB 3.2 Gen 2 ports to the Innuos Phoenix USB reclocker (FTA usb cable), then from reclocker to DAC (Danacable TruStream USB). As you probably know, the chosen USB port on a motherboard does matter.
At first, great review!
You said:
sinc-L: By far the most resolving filter on this list. It digs DEEP but remains very analog sounding. I could start to hear the individual members of the audience. Cymbals and hi-hats have more “clang” and material to them. There’s an organic weight and presence to all instruments and performers. Very transparent and separation is uncanny. Amazing filter if you enjoy an uncolored sound. Personally, I prefer a more warm-blooded tone.
Why didn’t you test sinc-L with LNS15?
And about DSD options, what is your opinion?
Keep in mind those filter impressions are BEFORE comparison to HMS. Everything is relative.
“I prefer a warm-blooded tone.” Just did another test (Alicia Key’s Falling and Woman’s Worth from Unplugged album) at 4x upscale. The tone on the HMS is far more accurate IMO. You hear more definitions from low-end as well. Cymbals have more metal to it and vocal overlays are sweeter. I still think the HMS performs better overall vs sinc-L and LNS15. The HQPlayer is smoother, zero grain…but doesn’t color the voices and instruments properly. It also doesn’t breathe into the room like the HMS. Also, the background of the HMS is quieter which helps with separation. I’ve stood 4 ft away while she was performing a few years back…the HMS reminds me of her stage presence…the HQPlayer does not.
I’ve replied to your earlier comment. I’ll have to test DSD files later.
Thank you so much!
I’m waiting your DSD tests…
There is a lot to unpack and a lot of value in this review. Unfortunately, it has been my experience that HQPlayer has the least to add to the Chord set of DACs because they effectively limit upscaling to PCM input and the TT Scaler is optimized for that. For those with ESS ES9038PRO or AKM AK4490 chipset-based DACs or R-2-R ladder Dacs (Holo Audio or Denafrips) using HQPlayer to upsample to DSD256 or DSD512 seems to offer a greater level of improvement than is possible using it to upsample PCM for Chord. That then begs the question: Is there a HQPlayer/DSD512 non Chord DAC combo that sounds as good or better tban the TTScaler/Dave combination at a similar price?
Which DACs would be good candidates for such an experiment? I’m curious and willing to hear them out.
A Chord TT Scaler/Dave versus Denafrips Terminator II + Gaia + HQPlayer comparison would be pretty interesting. My guess is that the Gaia might add some of the timbre accuracy you noticed in your Terminator review and the right HQP filter at DSD512 would add even more delicacy.
Hi – just saw this as I was writing an earlier post – I suspect you are right about Chord DAC’s not being the best fit, as I believe any DSD inut is turned to PCM and then back into DSD (no idea how that is done, but I’m think that is broadly right?). however, with my Chord Qutest I definitely get some benefit of using HQP in DSD mode, and so while its certainly possible that some DACS respond better to this, Chrid still seems to get some of the effect, and as a user who was considering adding an M-Scaler, a review looking at comparing DSD via HQP to the M-Scaler is very much of interest to me (as is whether other DAC’s respind better :-))
Can someone confirm whether HQP converts DSD to PCM first? I’ll try DSD via HQP the Chord DACs in the meantime.
I’ve added more info in “The Battle” section. TLDR the HMS is still better to my ears but that specific mode is better than PCM output.
Hi I realize this was a while ago but how are you connecting to your qutest? USB? I find with USB I can’t go above 2 or 4X upsampling without getting popping. Are you using the dual bnc inputs?
I no longer have the Denafrips but yeah, that would be interesting. I have not heard the Gaia DDC.
I think you should do a full review of HQPlayer and mscaler with the Schiit Yggdrasil DAC. As many have that DAC and would love to see how it performs under these conditions.
Great write up—really nicely done. Question: did you have the Phoenix USB reclocker in your chain for these tests? If you did, I assume it was before the M Scaler, and for HQ Player, right before the DAVE?
It was between the PC and DAC. FTA Callisto USB from PC to Phoenix and Danacable TruStream from Phoenix to DAC. I love this USB reclocker.
Great review Jay! I was considering using HQP but now not so much, you answered that question quite well… I have an HMS and Dave used via Roon and a couple of questions come to mind: How does the Phoenix USB reclocker with the HMS/Dave combo compare to using something like a Sonore opticalRendu (that’s what I use – fed via a SonicTransporter i9 server) which then sources the MScaler going to the Dave with Black Cat Tron BNC cables and a few ferrites. The optical Rendu is way better sounding than USB direct from the server to the HMS. Also, have you experimented with an LPS for Dave? I just got a Sean Jacobs DC4 LPS (it’s a kit so easy to install, but not cheap). Not a subtle change to say the least! Hopefully this LPS can be a potential future review by you. Again thanks for your great review!
I’m sure HQP will sell well regardless. For a non-hardware solution, it’s pretty nice. I honestly can’t tell you because I don’t have the opticalRendu or SonicTransporter. Black Cat also didn’t want to send in a BNC cable lol. I have not experimented with an LPS with the DAVE but I’ve heard one (A bespoke by Sean Jacobs). Crazy improvements in transparency, depth, and texture…but I felt like (the one I heard at least) it took away too much warmth. I don’t think it’s the DC4 LPS but I’ll have to check. $7,400 is quite hefty! lol. Maybe he’ll send one in for review………..
As for the Innuos Phoenix, once I heard it…it was difficult not to buy it. It wasn’t a planned purchase – but damn…it sounds good.
I agree with Nick’s comment below that the LPS (in our cases the Sean Jacobs DC4) is a huge improvement and doesn’t take way warmth but it doesn’t add any either, but in all other respects is crazy good. Perhaps you heard a DC3 Sean Jacobs or some other variant.
I agree with your comment in the review that it would be great if Dave could have a “warm” filter like the TT2 does.
I’ll likely be trying Nick’s Wave cables soon, perhaps that’s the added touch of warmth I’d like.
It would be great if at some point you could review the opticalRendu. It was a 100% positive improvement in my system (and I source the data to the opticalRendu directly from my SonicTransporter i9 server via optical fiber – it never goes out to the ethernet switch at all). Always something to try!
As Mr. Dolezalek notes “For those with ESS ES9038PRO or AKM AK4490 chipset-based DACs or R-2-R ladder Dacs (Holo Audio or Denafrips) using HQPlayer to upsample to DSD256 or DSD512 seems to offer a greater level of improvement”. I output HQ Player almost exclusively to DSD with HQPlayer. My modest Win10 PC comfortably runs HQPlayer with FLAC files output to DSD128 9or DSD256 with ASDMEC modulator, Sinc-L filter into a Singxer DDC, then I2S to an ESS-9038 based Matrrix X-Sabre DAC. The sound quality for my preference is far superior to HQ Player’s PCM to PCM upsampling, and I agree it would be terrific to hear your impressions of HQPlayer DSD output versus HMS with Chord DAC. Thanks again, Audi Bacon, for helping those of us without HMS to gain insight into it’s performance versus HQP Player. Great stuff!
Superior in what way? OK, I’ll start doing some DSD output tests. Also, are there any specific top-tier DACs that are popular with HQPlayer’s DSD upsampling?
Why not try this newest iFi Diablo DAC/Amp with so many great reviews?
I read a review comparing this $900 ancient Burr-Brown DAC chipset with $8,500 Sony: https://www.youtube.com/watch?v=3t5sO246_Ko and you can test with it PCM768 and DSD512 with HQPlayer!
So far, I’m getting a ton of dropouts with the EC modes. And it takes FOREVER to load lol
sinc-L and EC modulators are PC killers!
There is no capable PC to run sinc-L with EC modes DSD512!
{Think my previous comment did not show up, so I’ll try again here}. First: thanks for the detailed review. (1) Online reports indicate that HQP used OPTIMALLY beats Dave + MScaler. (2) A recommended optimal combo: the inexpensive RME ADI-2 DAC FS (version 2) (Jussi, the HQP developer, uses this too) connected via USB to the $800 M1 Mac Mini (8GB RAM is sufficient) hosting HQP desktop. Upsample to DSD256 using ASDM7EC modulator and poly-sinc-ext2 filter. You should get little to no latency with the M1 Mac Mini. This might save you a bunch of money in the long run and we might find your MScaler / Dave / TT2 in a ‘for sale’ listing online. Performance improves a bit when you use a good NAA in between the HQP and the DAC (like the ultraRendu or the SoTM SMS-200Ultra).
Sorry, meant to specify ASDM5EC modulator above.
Given that the use of wave BNC cables is considered to be mandatory for connecting the M-scalar; the value proposition is more like $260 Vs $7,000
True true…but if it’s not what you’re looking for…
Great review Nick! I’ve been using hqplayer for many years now and agree with most things.
I like 768khz upsampling the most which is 16x since but since LCD4 are so thick to begin with it doesn’t end up sounding too thin. I also cheat and use – 1dbfs which to me sounds much better than – 3dbfs, much less constriction and as long as my limiter doesn’t hit, I know I’m not clipping. It also helps the LCD4 soundstage a lot, combined with no dither and poly sinc LP, gives me the best balance.
Sinc L is incredible, it also takes me 6 seconds to run so I don’t think the processing time varies much.
The r2r dacs I have, have reacted much better to upsampling than DS ones. I prefer to not turn on multi core dsp as it smears highs and removes lower fundamentals a bit at the cost of soundtsage, but I’m using an old i5 cpu. Also in your asio manager you can turn off (safe mode) which elevates the fidelity even higher, changing the buffer there changes the energy in the highs.
You can also use parametric eq on hqplayer through the matrix pipeline. The manual also states some dithers are better suited to different levels of upsampling.
Yeah, if I had an LCD-4, I would go 16x as well lol. No shortage of warmth there. Good info. Hm, how do you disable safe mode with a Chord DAC?
Addendum regarding DSD posted in “The Battle” section.
Fantastic that you tried the DSD option – I actually agree with you in respect of the ADMS7 shaper you used – I personally found it not so much different to the PCM settings, and seeing as my Chord DAC is DoP I left it in PCM mode – until i tried the EC shapers – these were, for me, a big step up, even though I had to reduce from 256 to 128 – despite the power of your rig, you might need to step back from 512 and try it in 256, but that should still be much more to your tastes than 512 using a non-EC shaper . I do have 5 secs or so wait at the start of an LP, and that’s about my limit. Might also be of interest that I am using a small piece of cmd line SW to use Jriver as a frontend on my collection, and clicking on any track from an LP in jriver pushes the whole album to HQP and starts playing from your selected track – really nice functionality
Hi Jay, hope you are well.
On HQP I broadly agree with with your conclusions in all respects and for me it is not viable replacement for the Mscaler, especially with my WAVE cables on the Mscaler if I may be so bold! HQP is rather like listening to a TT2, it sounds good in isolation but if one hears a Dave after the TT2 then the TT2 deficiencies become obvious and so it is with HQP when one listens to a ‘sorted’ Mscaler compared to HQP. By ‘sorted’ I mean using a different power supply and of course sorting the BNC cables.
The DC4 has been mentioned. I have the Sean Jacobs DC4 power supply with my Dave and prior to that I had the DC3. As you say, there are crazy improvements in transparency, depth and texture by adding the DC4 but I heard no loss of warmth. In fact to my ears I hear less harshness. I would never have described Dave as being harsh until I swopped back to the stock Dave after using the DC4 powered Dave. When I was assessing the DC4 effectiveness with the Dave I was lucky at that time because I had a second stock Dave so I could do more or less instant switching between them. The price was steep when I bought my DC4 (I had the first production DC4 that Sean made) and it is steeper now but everyone who has bought a DC4 even at the current price has come back to me and thanked me for suggesting the stunning upgrade to Dave.
Just back on the subject of HQP, I do sometimes use it with all upscaling turned off (ie set to ‘none’ for Filter and Noise Shaper) and then I find the HQPD app + HQP + HQP NAA player to be a well executed player to rival Squeezelite etc. Note that I do not use Roon + HQP because to my ears Roon erodes the sound quality to a point where it means that I do not enjoy listening to my system.
Finaly, there have not been any product announcements from Chord for a long time, probably due to Covid. To my eyes there is a glaring gap in the line up previously occupied by the Blu Mk2. It would not surprise me to see a Choral range Mscaler to compliment the Dave. If so I am guessing that will far outstrip the current Hugo Mscaler and set the bar even higher for rival upscaling products.
Keep up the good work. Nick.
Loving your STREAM cable! It has my highest recommendation. OK…I really need to hear the DC4. I’ve only heard great things.
Sorry, I’m not on the forums much. How exactly do you go about setting up that trio of HQP? Have you tried the latest Roon?
Hi Nick, Your experience with the Sean Jacobs DC4 matches mine. I was actually hoping that it would give me some added warmth (it didn’t) but it didn’t take any warmth away – and it did add a very large improvement in virtually every other respect, I can’t go back to not using the DC4 at this point!
I find it interesting that Jay prefers Roon and you prefer not-Roon. I use Roon with my SonicTransporter i9 (and optical Rendu and MScaler also with LPS) but I haven’t done much exploring of non-Roon sourcing. I know folks vary on this issue, but the small group of folks at AudiophileStyle who are really going to the extreme with Dave mods, servers etc. (they all seem to prefer the Taiki Extreme) agree with you in not preferring Roon. I do wonder though if Roon v1.8 improved things (and might Roon v2.0 not too distant)? Tradeoffs…
I really should try your Wave Storm Reference cables soon and compare to my Black Cat Tron (as we’ve discussed before). I have a strong feeling that I’ll like the weight and tonal balance of your cables better than my Trons even if the Trons may (not sure) provide better detail and space. I tend to align with Jay in that the tonal color is super important to me and I need more warmth. It would be great if the new Dave (if there is one or maybe a FW update) could have a “warm” filter to use like the TT2 does. Rob Watts is working on a 16M tap “MScaler” and I hope it comes out soon! Then again, a new Dave v2 with a great LPS would likely be even better but per Rob is not in the cards anytime soon…
Please give xxhighend a try. It has a custom interpolator which he calls “arc prediction” and it sounds fantastic. (Have tried hqplayer and wasn’t quite impressed, but I expected it considering the linear structure of the filters). Xxhe is a bit hard to set up but worth the effort imo.
Also, make sure you set windows/dac volume to 100% and use xxhe to control the volume.
I’ve been comparing the M Scaler with HQPlayer recently. Enjoyed your write up and the settings explanations etc.
Connected directly to the DAC, I prefer the M Scaler… for mostly all the reasons you give.
However, using HQPlayer with OpticalRendu with the PC connected to the network via optical as well as the OpticalRendu it was a far closer battle. I found I stayed on NS9 with Sinc-M at full upscaling. Imaging and tonal detail was superb with a lot of body. Via OR, I found the LNS15 to be too heavy, thick.
System. Roon NUC – Optical Rendu – MScaler – TT2 – Etude with TQ Ultra Silver XLR – PMC 25.22 + LPS’s for OR and network gear.
Would you try also DSD512 with sinc-M, sinc-L & poly-sinc-ext2 also with OpticalRendu?
I would but I don’t have any DSD content and I haven’t been able to find any links to any.
Would be high-res FLAC files upsampled to DSD512!
Not sure if you mean upsample the file beforehand or use HQPlayer.. I don’t have any software to convert the file itself so…
I set HQPlayer bit rate to 48k 512 with all the oversampling options requested plus Closed Form 16m and the DSD256 file I was using worked fine on all of them.
Computer used for ref is a Mac min 2018, 3.2 GHz 6-Core Intel Core i7, 32gb ram and Vega 56 eGPU.
Jay, Great review but any idea how it compares to the Directstream? I currently have a Directstream and heard a friends Qutest. Really liked it and was thinking about trading my Directstream in on a TT2. Looking for a little more “Bite” in my listening. Thoughts? Thanks.
I think you’d be pretty happy with a Qutest…unless you hear the TT 2 lol. I’ll be comparing them to the RME soon.
I really liked this review. I look forward to the best DSD settings!
SRC-DX dual BNC into a TT2/ Dave gets you M-Scaler level (or arguably better) performance as you bypass the relatively noisier Amanero USB connection and take advantage of the ideal dual BNC connection. Suddenly the M-Scaler no longer has monopoly on the use of dual BNC in Chord DACs.
Can you play dsd512 with sinc-l and 7ec? No one knows if this setting will actually work with 3090.
Interesting review and to read the numerous opinions among the comments. Thanks to my work as an electrical engineer, my rig has been overhauled to be as dead silent as I guess possible in a domestic environment. The vast majority of gear, even high-end, break the rules of good engineering and cannot suppress the noise from the invironment the gear is normally working, the audiophile’s home. We had a battle with the Hugo and the MScaler at a friends place just a couple of weeks ago. The rig subject for testing was not of the fanciest level, but well the Rotel Michi pre and poweramp.
The conclusion from the test was that the Chord digital devices sounded harsch, overemphasising the high mid and top, with clear resonces ringing and listening to some highly processed gothic metal it was just to shut down and choose something simpler to play. At my home, I use a Roon ROCK -> HQPlayer Embedded. I output oversampling from any format/depth or speed to DSD256 DoP -> USB -> RME ADI-2 fs DAC for about USD 1.000. Poweramp is Benchmark ABH-2 and speakers Marten Parker Trio. Using the HQPe via ALSA backend, I listen to numerous genres and I have not been able to balance the insanly poor post process of today, also when remastering old albums … until now. HQPe sort this out. It is able to in almost all cases playback the music I like, irrespective of genre. The ability to penetrate all tiny details in a classical arrangement as well as contemporary music is again making me interersted in my album collection. Not to mention how easy it is to detect the differences between streaming services. I can agree that the fight to get into the HQP world is a painstacking investment, but so far, with a lot of help from the very resourceful community of Roon, it is worthwhile. I would never trade my chain of playback to anything else, unless Anti-Pode new server/streamer/player units throw me of the chart. I am anxiously waiting for my turn to borrow one.
Hi Jay, I have a question can you please email me. Thanks Brad
Thank you for this great review! Trying some of your recommended settings already. Maybe worth to not, when you use e.g. in Roon a playlist or so, you do not have this 30 seconds between tracks. Tracks are prepared in the background. Anyway thank you agin 🙂
You’re very welcome!
you guys are hilarious – you should never use FLAC – it sounds worse than Wav…. PCM wave to DSD as high as you can go or 1xDSD to 4X or 8X DSD is where it is at. FLAC IS CRAP
How is FLAC crap? Convert a WAV file to FLAC and then back to WAV and the resulted WAV file will have the same checksum as the original WAV. FLAC is just a container, with the option of lossless compression.
Jay, HQPlayer 4 Embedded PCM setting have 1x filter and Nx Filter with Dither, the combo you given has only one filter and one dither, should the filter be the same for both 1x and Nx? and what about SDM only has 1x oversampling and Nx oversampling without dither, what is your reference for the SDM? thanks.
What about hardware implementation at source level, tonal differences might be due to outputting from the M-Scaler digital outs (most likely very well designed) and whatever digital source component you used with HQPlayer.
Theoretically the math and processing power behind MQP is simply more powerfull. So there might be other reasons.
Have tried them and when the rig is silent enough, there will be no need for long sessions to establish that HQPe have the potential of the utmost silence and correction of todays most torture music worshipping of the limiter and compressor instead of skilled mix and gentle mastering. But the ethernet network, the mains, and each individual device ground must be carefully scrutinized and corrected (a challanging project as there are numerous devices today lacking ground). The sound degrading high frequency multi source interference is by far the most troublesome area, comparing to any re-sampling issues. MScaler does not offer even close to the resample power of a computer like the one in this test (but carefully treated to isolate the pc noise, needed off course), and turbo boosted with Cuda core off-load, the modulator ASDM7ECv2 and filter poly-sinc-gauss-xla, in my case to DSD256 DoP, HQPe is so far unparalleled. Even with my cheap RME DAC. Tonality, real-life resemblance, imaging, transients, phrasing, everything starts to sound as it did while being in the studio, before over-ambitious post-process mastering.