More comparisons again tonight. Still using the default TPDF shaper for PCM and haven't played with those options yet. I still tend to like PCM even though a DSD signal is supposed to have a shorter path out of the DAC chips of the Gustard A26. It is remarkable how different the various filters of HQPlayer sound. Switching onboard filter option in DAC menus make very subtle changes which often border on inaudible. The choices in HQP seem to make a much more noticeable difference. I did try the recommended Closed Form M but found that the Poly-Synch-Gauss-Long that came preset with my install is still my favorite. It is the widest, deepest, blackest, with the best fine resolution of reverberant tails. But it can sound a bit lean and remote. The Closed Form M is thicker and meatier but gives up too much width and fine detail. The P-S-G-Hires-MP is an interesting option It is fuller and closer than the PSGLong with a nice sense of ease. It is more like the SDM with the PSGLong filter and ASDM modulator but they are not as wide or deep as the LAN PCM 768 Long.
Latest Thread Images
Featured Sponsor Listings
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
HQPlayer Impressions and Settings Rolling Thread
L@eric
New Head-Fier
- Joined
- Sep 29, 2009
- Posts
- 33
- Likes
- 18
update Ubuntu Server 22.04 (Jammy Jellyfish) + HQPlayer Embedded 5.6 as follow
sudo dpkg -r hqplayerd
sudo apt autoremove
wget https://www.signalyst.eu/bins/hqplayerd/jammy/hqplayerd_5.6.0-17_amd64.deb
sudo dpkg -i hqplayerd_5.6.0-17_amd64.deb
sudo apt install -f
sudo hqplayerd -s hqplayer password
sudo systemctl enable hqplayerd.service
sudo systemctl start hqplayerd.service
Remark : I assume your HQPe server can execute avx2
sudo dpkg -r hqplayerd
sudo apt autoremove
wget https://www.signalyst.eu/bins/hqplayerd/jammy/hqplayerd_5.6.0-17_amd64.deb
sudo dpkg -i hqplayerd_5.6.0-17_amd64.deb
sudo apt install -f
sudo hqplayerd -s hqplayer password
sudo systemctl enable hqplayerd.service
sudo systemctl start hqplayerd.service
New to HQP, I have 5.5 embedded deployed and functioning but would like to update to 5.6. I'm struggling to find any documentation on the process to get to 5.6, can anyone point me in the direction?
Remark : I assume your HQPe server can execute avx2
dericchan1
500+ Head-Fier
Been listening to the dsc2 every night since Friday. Switched to the Cyan 2 tonight to compare.the output stage is.
Imagine HQPlayer implemented on FPGA inside the DAC for digital filtering and volume control. It won't be anywhere near as flexible or high fidelity as HQPlayer, but, the general idea is the same.
After the FPGA, all audio is modulated back to true 1-bit, 128fs. (Maybe 1 bit, 256fs in later models.. I am not sure about this). The output stage is similar to DSC DAC with multiple element, unary coded converter. The analog output uses output transformers.
So it is similar in some ways, quite different in others.
I do know they make constant 'improvements' via firmware. The firmware updates all seem to update the modulator parameters and maybe a few other things.
With hqplayer set to dsd512x48 7EC super
Cyan 2 sounds great like a great sounding dac should.
The DSc2 just sounds like a great sounding turntable.

Last edited:
plumpudding2
500+ Head-Fier
Trying to enable dac correction with Poly-sinc-gauss-xla, DSD512, ASDM7EC-light 512+fs my GPU memory usage goes to 11/11 GB and CPU ram usage goes to 32/32 GB, causing my computer to slow down to the point everything starts to freeze.
Switching to PCM x32 rate causes 4 GB of GPU VRAM and 1.5 GB of CPU RAM. More manageable.
Determined to make DSD work i switched to DSD256, ASDM7EC-super 512+fs, poly-sinc-gauss-short. Now we're at 11GB GPU VRAM and 13 GB CPU RAM, but it works! fabulous sound
For reference I have an intel 8086k + Nvidia 2080 Ti
EDIT: gauss-XLA also works at 256, so the memory requirements seem to be based purely on output rate and not filter selection
Switching to PCM x32 rate causes 4 GB of GPU VRAM and 1.5 GB of CPU RAM. More manageable.
Determined to make DSD work i switched to DSD256, ASDM7EC-super 512+fs, poly-sinc-gauss-short. Now we're at 11GB GPU VRAM and 13 GB CPU RAM, but it works! fabulous sound

For reference I have an intel 8086k + Nvidia 2080 Ti
EDIT: gauss-XLA also works at 256, so the memory requirements seem to be based purely on output rate and not filter selection

Last edited:
Thanks, that is helpful to understand. Does this principle apply whether one uses an NOS DAC or otherwise?If it is correcting for the DAC's analog filter it may not need to.
Afaik the holo analog filter is in the output stage itself and applies to both DSD and PCM
plumpudding2
500+ Head-Fier
This might be pure placebo but I think for the near future i'm going to stick with DSD256, ASDM7EC-super 512+fs + dac correction.
If I had to describe it I would say there's a definite increase in snappiness and weight.
Thanks to the measurements posted earlier in this thread it's comforting to know that technically DSD256 is completely fine, DSD noise at 20khz is -220 db instead of -260 db. No big deal as it is still miles away from the analog noise floor of -140db.
If I had to describe it I would say there's a definite increase in snappiness and weight.
Thanks to the measurements posted earlier in this thread it's comforting to know that technically DSD256 is completely fine, DSD noise at 20khz is -220 db instead of -260 db. No big deal as it is still miles away from the analog noise floor of -140db.
Bleuciel
New Head-Fier
Hi, I'm using Hqplayer for some time now, mostly upsampling with PCM to 764k. Recently I got the Topping DX9 and I'm experimenting with DSD512 upsampling. The sound is good but it is difficult to run it, lol. I have to enable Multicore DSP and CUDA offload or it won't run smoothly. Even then I can't use the most expensive filters on DSD512. I have 48 GB RAM, 12 GB GPU with a Ryzen 7 3700X.
With the PCM upsampling I can multitask on my PC which is a big plus. With The DSD upsampling I can't use the GPU for another task, so I use it mostly while internet and document browsing.
My favorite DSD filter is the closed-form-16M, with a more analog sound. Close second is poly-sinc-gauss-long and poly-sinc-gauss-xl. On DSD512 I use the AMSDM7 512+fs. The AMSDM7EC 512+fs and other EC modulators have stops on my PC on DSD512 and I can't really use it.
I'll try with the ASDM7EC-super 512+fs modulator on DSD256 as it won't run well on DSD512.
With the PCM upsampling I can multitask on my PC which is a big plus. With The DSD upsampling I can't use the GPU for another task, so I use it mostly while internet and document browsing.
My favorite DSD filter is the closed-form-16M, with a more analog sound. Close second is poly-sinc-gauss-long and poly-sinc-gauss-xl. On DSD512 I use the AMSDM7 512+fs. The AMSDM7EC 512+fs and other EC modulators have stops on my PC on DSD512 and I can't really use it.
I'll try with the ASDM7EC-super 512+fs modulator on DSD256 as it won't run well on DSD512.
Bleuciel
New Head-Fier
I recently got into virtual cables to be able to use Hqplayer with other sources than Roon. A guide on Stereonet is available. I can now use Hqplayer with YouTube, Spotify and all sources so it's quite useful. Note there is still a 1-2 seconds delay with the sound output on PCM upsampling, longer with DSD.
Last edited:
markkr
Headphoneus Supremus
Thanks VERY MUCH!update Ubuntu Server 22.04 (Jammy Jellyfish) + HQPlayer Embedded 5.6 as follow
sudo dpkg -r hqplayerd
sudo apt autoremove
wget https://www.signalyst.eu/bins/hqplayerd/jammy/hqplayerd_5.6.0-17_amd64.deb
sudo dpkg -i hqplayerd_5.6.0-17_amd64.deb
sudo apt install -f
sudo hqplayerd -s hqplayer password
sudo systemctl enable hqplayerd.service
sudo systemctl start hqplayerd.service
Remark : I assume your HQPe server can execute avx2
bogi
100+ Head-Fier
- Joined
- Mar 16, 2013
- Posts
- 402
- Likes
- 434
That's why these options are present.I have to enable Multicore DSP and CUDA offload or it won't run smoothly.

Those super long filters are not necessary to fully enjoy HQPlayer. There is no such equation that longer the filter, better the sound. In frequency domain it may be the case but not in time domain, which is about transients.Even then I can't use the most expensive filters on DSD512.
On Intel platform it helps to use integrated graphics for other apps. In your case it could help to add yet one inexpensive GPU. Win11 allows to define which GPU an application should run on.With The DSD upsampling I can't use the GPU for another task, so I use it mostly while internet and document browsing.
Yes, just try it. It is OK at DSD256, it would have drawback only with source content at sample rates >= 192k with transients exceeding 50k.I'll try with the ASDM7EC-super 512+fs modulator on DSD256 as it won't run well on DSD512.
bogi
100+ Head-Fier
- Joined
- Mar 16, 2013
- Posts
- 402
- Likes
- 434
Many video players do support youtube links and allow to set video delay against audio. That way you can synchronize video with audio even when doing PCM to DSD upsampling.Note there is still a 1-2 seconds delay with the sound output on PCM upsampling, longer with DSD.
I am using PotPlayer on Windows for this.
MLGrado
500+ Head-Fier
Been listening to the dsc2 every night since Friday. Switched to the Cyan 2 tonight to compare.
With hqplayer set to dsd512x48 7EC super
Cyan 2 sounds great like a great sounding dac should.
The DSc2 just sounds like a great sounding turntable.
![]()
yeah... i can totally understand that...
but, it has superior dynamic range, frequency response, extremely low jitter (compared to wow/flutter on turntable), and of course no rumble, no pops and ticks etc, compared to vinyl haha.
But agreed it sounds very, very analog.
I am hesitant to post any of my measurements because when I took them, there was an impedance mismatch between the DSC2 and the E1DA ADC. That has since been rectified with the addition of E1DA's 'scaler' device. I am all set to take new measurements, so take these below with a grain of salt. They may be better than this.
However, I decided to reveal them with full knowledge I will be re-measuring them soon, and will post the updates in this thread.
--------------------------------------------
EARLY DSC2 MEASUREMENTS



The graph above is where you can see the output transformer signature. SINAD stays at 90db and above from approx. 57hz to 18khz.
preferably for a DAC, I would like to see SINAD at closer to 100db. (And of course the latest state of the art DACs are way above that). But there are some really fine sounding DACs that have their SINAD in the mid to high 90's. For instance, the Burr-Brown DSD1793 (per single chip) has a best case scenario of 0.001% THD+N. Which is exactly 100db SINAD. And I find it sounds just fine. Actually, my reference DAC is the iDSD PRO.
But what you CAN see is the very low SINAD as frequency drops below approx. 57hz.. Down to 76db at 20hz. Also, at 20khz its as low as 82db.
IMO, this must be where the analog sound comes in. It actually reminds me a bit of tube sound, which will and most often does have SINAD even worse than this.
I was actually quite impressed with the DSC2 and how long it stayed over 90db SINAD through most of the audible range, and I don't hear any issues personally in the bass.
It always sounds full and tight enough.
I placed an order today for the original version of the PSAudio DirectStream DAC. It is NO measurements darling. It uses similar tech to the DSC2, but instead of the superior filters on computer hardware, they are on a FPGA. Output is also via transformers. I am curious as to how it performs compared to the DSC2. It is RIDICULOUSLY expensive for very similar implementation. The 'Audiophile' markup. 'Magazine Engineers' seem to hate it; listeners of course seem to love it.
Myself? I am just too curious for my own good. And for the good of my wallet!
dericchan1
500+ Head-Fier
Much appreciated the measurements above, I might be mistaken, I wonder if the tube sound actually coming from the fact that (from the last graph) SINAD drops significantly from 15khz to 20khz from 98db to 82db - meaning there's quite a bit of distortion in the HF?yeah... i can totally understand that...
but, it has superior dynamic range, frequency response, extremely low jitter (compared to wow/flutter on turntable), and of course no rumble, no pops and ticks etc, compared to vinyl haha.
But agreed it sounds very, very analog.
I am hesitant to post any of my measurements because when I took them, there was an impedance mismatch between the DSC2 and the E1DA ADC. That has since been rectified with the addition of E1DA's 'scaler' device. I am all set to take new measurements, so take these below with a grain of salt. They may be better than this.
However, I decided to reveal them with full knowledge I will be re-measuring them soon, and will post the updates in this thread.
--------------------------------------------
EARLY DSC2 MEASUREMENTS
The graph above is where you can see the output transformer signature. SINAD stays at 90db and above from approx. 57hz to 18khz.
preferably for a DAC, I would like to see SINAD at closer to 100db. (And of course the latest state of the art DACs are way above that). But there are some really fine sounding DACs that have their SINAD in the mid to high 90's. For instance, the Burr-Brown DSD1793 (per single chip) has a best case scenario of 0.001% THD+N. Which is exactly 100db SINAD. And I find it sounds just fine. Actually, my reference DAC is the iDSD PRO.
But what you CAN see is the very low SINAD as frequency drops below approx. 57hz.. Down to 76db at 20hz. Also, at 20khz its as low as 82db.
IMO, this must be where the analog sound comes in. It actually reminds me a bit of tube sound, which will and most often does have SINAD even worse than this.
I was actually quite impressed with the DSC2 and how long it stayed over 90db SINAD through most of the audible range, and I don't hear any issues personally in the bass.
It always sounds full and tight enough.
I placed an order today for the original version of the PSAudio DirectStream DAC. It is NO measurements darling. It uses similar tech to the DSC2, but instead of the superior filters on computer hardware, they are on a FPGA. Output is also via transformers. I am curious as to how it performs compared to the DSC2. It is RIDICULOUSLY expensive for very similar implementation. The 'Audiophile' markup. 'Magazine Engineers' seem to hate it; listeners of course seem to love it.
Myself? I am just too curious for my own good. And for the good of my wallet!
MLGrado
500+ Head-Fier
Much appreciated the measurements above, I might be mistaken, I wonder if the tube sound actually coming from the fact that (from the last graph) SINAD drops significantly from 15khz to 20khz from 98db to 82db - meaning there's quite a bit of distortion in the HF?
yeah, very possible.
I should run a test with the fundamental 5khz or higher to see what the harmonic distortion looks like up there.
BShaw
1000+ Head-Fier
Since I'm waiting for my Cyan 2 to arrive, I've had a lot of time to think. I ordered the Cyan to replace my Gustard R26 in order to run DSD from HQP, thinking the Holo looks like a great "DSD machine" and probably a good value as well. Previously I'd been lusting for a Spring KTE as what I'd have to stretch to in order to appreciably outpace the R26 (and play DSD files).
Then came the Laiv Harmony. I first heard about it on John Darko's podcast with Srajan Eiben and was intrigued, and within two days it was everywhere with people inventing new superlatives to try to express their joy. Damn, $500 cheaper than the Spring and quite possibly a step beyond, maybe two steps.
This got me thinking, I wonder how the Laiv playing 768 PCM would compare to the little Cyan 2 running DSD? After all I've heard about DSD being clearly more organic, realistic, analogue, etc, I started to question everything. Go figure.
So now I want to put it out there and I'm really interested in what people have to say. I wonder first of all if there's anyone who thinks the DSD vs PCM thing might be a bit hyped (or not), and also if you think a DAC of the Laiv's class might overall be a better choice even with just PCM, because......it's just a better DAC?
I know this is pure speculation, but that's all I'm going to have to go by between now and the time I possibly cancel my Cyan order and send some money to Singapore.
I'm really looking forward to what some of you guys might have to say, there's a lot of knowledge and experience in this thread.
Then came the Laiv Harmony. I first heard about it on John Darko's podcast with Srajan Eiben and was intrigued, and within two days it was everywhere with people inventing new superlatives to try to express their joy. Damn, $500 cheaper than the Spring and quite possibly a step beyond, maybe two steps.
This got me thinking, I wonder how the Laiv playing 768 PCM would compare to the little Cyan 2 running DSD? After all I've heard about DSD being clearly more organic, realistic, analogue, etc, I started to question everything. Go figure.
So now I want to put it out there and I'm really interested in what people have to say. I wonder first of all if there's anyone who thinks the DSD vs PCM thing might be a bit hyped (or not), and also if you think a DAC of the Laiv's class might overall be a better choice even with just PCM, because......it's just a better DAC?
I know this is pure speculation, but that's all I'm going to have to go by between now and the time I possibly cancel my Cyan order and send some money to Singapore.
I'm really looking forward to what some of you guys might have to say, there's a lot of knowledge and experience in this thread.