@andybob @jussi_laako Many thanks, noted. Especially “set and forget” is very encouraging as the starting point.
HQP Desktop experience encouraged me to experiment with HQP Embedded, the question now - can I tame it all right !? hehe 
I just looked at the time when this thread has been started. OMG, looks like I’m 6 years late to the party… hehehe! In meanwhile over the past few days I think I found my Holy Grail in solo piano sonic reproduction.
For the listening I took the following pieces:
- Keith Jarret, The Köln Concert (SACD, CD and Qobuz 96/24) - I guess this is his most famous live concert work
- Keith Jarret, The Budapest Concert (Qobuz 48/24)
- Keith Jarret, Radiance, 44.1/16
- Glenn Gould, Bach: The Goldberg Variations (SACD and Qobuz 44.1/24) - this has intentially a bit clavier sound
- Philip Glass, Solo Piano, 44.1/16
- Lubomyr Melnyk, Fallen Trees, 44.1/24
- Lubomyr Melnyk, Illirion, 44.1/16
- Lubomyr Melnyk, Three Solo Pieces, Qobuz 44.1/24
- Group 180, Piano Phase (Steve Reich), 96/24
And the most peculiar piano piece ever which can only be pefromed by two people in this world, the master and his student.
- La Monte Young, The Well-Tuned Piano, 44.1/16
The winner combination limited by the processing capacity of my hardvare (fanless i7 NUC) is:
poly-sinc-short-mp + ASDM7EC @ DSD 128
For solo piano, which was my original intention, the realism of reproduction is mind breaking!
I suspect that Jarrett’s Köln mastering is already with artefacts, so plain PCM with “no filters” made me very much concened that something was wrong with my DAC. I did try on two other chips ESS and AKM, to peace in mind, the results were the same - quite unnatural piano. Nonetheless, HQP manages to bring more life and naturalism in perfomance. So obvious that blind test is no concern here at all…
The rest of the list was just the confirmation that this filter was a good choice, until the works of Lubomyr Melnyk. Melnyk is famous for producing the wall of grand piano sound with often long open pedal. In all my previous encounters with his music I never managed to finish listening any of his works, basically it was a head ringing torchure. Until… poly-sinc-short-mp changed things completelly. Now, instad of painful ringing, is the mellow and deep tembre of the grand piano in low registeres, with clear (ok, clear as much as it was recoded) highs reproduction, all without that hammering ringing in your brain. While at start I thought I would just give it a try, not expecting anything extraordinary, I ended up listening all the 3 works in one go, coming back to Fallen Trees to hear it again.
The work of La Monte Young has acquired the realism, and now makes much more sense with this new way of reproduction.
The other filters for solo piano I’ve tried were as suggested above - “ext2” and “poly-sinc-gauss”, while I totally agree these are the “step up” from whatever any default player would offer in solo piano excercise, the “poly-sinc-short-mp” was like just made for this purpose.
Interestingly enough, the “poly-sinc-xtr-short-mp” was not a good match for just these works. However, when listening to Keith Jarrett Qartet, “My song”, which is ermmm… jazz quarted - xtr was shining !!!
So hear come few questions, if I may. Although I’ve tried many of the possbile combinations that my system would digest, I kept coming back to ASDM7EC @ 128.
-
Is there a hirarchy of combinations of Modulators + SDM rates? In other words, would ASDM7 @ 512 be superior to ASDM7EC @ 128 per design?
-
And would (per design) ASDM5EC @ 256 be superior to ASDM7EC @ 128 ?
4 Likes
Yes, but:
- you would need a computer with single core performance of about 10 GHz to run 7EC at DSD 512 and they don’t exist outside of liquid helium baths in research labs yet; and
- there is a measure of ‘diminishing return’. The step from DSD 64 to DSD 128 is the most significant;
- the trade off between 7EC/5EC and 128/256 DSD requires listening to determine which is better for a given task. 7EC has finer detail, DSD 256 finer dynamics.
Glad to hear you have found combinations that suit you. Short mp is optimised in the time domain so is the classic choice for transient reproduction.
2 Likes
Not that way directly. Although you get improvement over increasing rate. It also depends on the DAC design.
Likely yes. For some converters and associated equipment ASDM5EC may be better choice overall than ASDM7EC.
2 Likes
And which one would you recommend for the nagra hd x (Andreas Koch design)?
ASDM7EC at DSD256, certainly.
hey all! Listening through a Focusrite 2i2 3rd gen Audio interface. I know it’s is not suitable, but all I got at the moment where I’m living. It can do up to 192k. Any filter recommendations you guys got for me? Currently running with NS1 and sinc-M!
Actually it’s a good interface !
Focusrite Scarlett 2i2 Audio Interface Gen 3 Review | Audio Science Review (ASR) Forum
I have it’s bigger brother 18i20 Gen3. Only weakness (to me) is headphone output but I don’t use it for that.
I use NS9 with everything upsampled to PCM192kHz . TPDF dither should be fine too.
poly-sinc-gauss and ext2 are nice
For upsampling to 192kHz I would suggest something like poly-sinc-ext2 or sinc-S with NS9 or TPDF as well.
I know sinc-L is non-apodizing (apodizing can be a good thing).
But just revisiting sinc-L with my Chord Hugo2, just wow ! (PCM705/768, LNS15).
There is something different about transients with sinc-L at 2M taps ! The better transients likely offsets the non-apodizing and extremely long pre-ringing feature ! 
With my RME ADI-2 this is too bright and fatiguing with extended listening. Not the case with the Hugo2.
The way I gauge filters is, does it make me want to turn volume up or down after an hour. sinc-L and Hugo2 makes me want to keep turning up ! No listening fatigue.
Transients on this - wowza
Currently using a M1 Mac Mini with Holo May.
I like both of these:
- PCM 768KHz with LNS15 / Sinc-L
- DSD256 with ASDM7EC / poly-sinc-ext2
Has anyone succeeded running Sinc-L with DSD256 / ASDM7EC?
The new M1 Max chip seems like exciting news for all HQPlayer aficionados!
Tried the new ASDM5ECv2 @ DSD128 and poly-sinc-short-mp-2s on my i5 8gb RAM laptop - 4.26.0 HQPlayer Embedded - Ubuntu 20.04 server, a bit more CPU juice is needed:
Same song and output using ASDM5EC:
Too early to tell if 5ECv2 is any better, thanks for the great work Jussi.
How is DSD256 with ASDM7EC / poly-sinc-gauss-xla with the M1 Mac Mini?
I was hoping they would release an M1 Pro or M1 Max for the Mac Mini the other day but no dice so I’m holding off on buying one for now. (but still kind of on the fence as it would still be an improvement over my current setup). I can’t justify buying a MBP just to be a Roon Core / run HQP (I mean I can but I definitely should not) 
Unfortunately the performance of M1 is not sufficient for this combination. I made two charts for your reference. The first chart is ASDM5EC with filters upsampling to DSD256 or DSD256x48.
The second chart is ASDM7EC with filters upsampling to DSD256 or DSD256x48. I think this is good for other platforms’ performance comparison and evaluation.
*My Mac mini M1 was configured as 8GB RAM and 256GB SSD.
13 Likes
Wow, this is very comprehensive, thank you for putting it together. I’m sure others will find it useful as well.
In my current setup I am running DSD256 with ASDM7 (non-EC) / poly-sinc-gauss-long without any issues, If I really push the threshold I can do DSD256 with ASDM5EC / poly-since-gauss. It sounds like the M1 Mac Mini would be an improvement, but not as much as I was hoping. My 4th Generation Haswell (4.0 GHz Core i7-4790K) from my Late 2014 Retina 5K iMac is holding up surprisingly well.
I’m curious to see what the new M1 Pro or M1 Max is capable of, I may be looking to get a MBP after all.
My point of view is I would not consider the system that cannot meet the poly-sinc-ext2 + ASDM7EC and upsamples to DSD256 or DSD256x48 for red book. It’s the bottom line of modern computer we’re looking for.
Nice work there 
Please recommend a ‘warm but detailed’ HQP filter.
DAC: Holo May L2
I recently changed my preamp from the McIntosh C2600 (tube preamp) to the Holo Audio Serene (SS preamp). Now, with my current speakers (Focal Sopra 3), I find some tracks to be a bit too bright.
Currently, I upsample to 768Khz PCM using LNS15 / Sinc-L or to DSD256 using ASDM7EC / poly-sinc-ext2 (with a slight preference towards the PCM config).
Thanks.
I’d suggest giving closed form a try.
Have you tried ext2 with PCM ~1.5MHz ?
Set output DAC bit depth to 18 or 20 in HQPlayer