Jump to content
IGNORED

HQPlayer - Favorite settings?


miguelito

Recommended Posts

The MQA thread made me think that all this chatter about fixing the ADC and DAC errors can, to a great extent, be accomplished as an on-the-fly process at the playback end. I expect the majority of MQA content will be post-processed digital files from the final files sent to CD production.

 

So I have two questions for Miska et al:

 

1- This de-blurring effect people are talking about - can it be implemented as a post-process filter?

 

2- In light of all the chatter of improving whatever issues are present in the input file, does this mean that upsampling might get a new think-through ie go past interpolation and further into "fixing" the original data?

 

3- The argument about tailoring the fixes to the ADC and DAC used could obviously be incorporated into settings of an on-the-fly filter - is this a possibility? (*)

 

Thx.

 

(*) iZotope in Audirvana allows you to tune some output parameters. I imagine different DACs would sound best with their tailored settings - I haven't investigated this.

NUC10i7 + Roon ROCK > dCS Rossini APEX DAC + dCS Rossini Master Clock 

SME 20/3 + SME V + Dynavector XV-1s or ANUK IO Gold > vdH The Grail or Kondo KSL-SFz + ANK L3 Phono 

Audio Note Kondo Ongaku > Avantgarde Duo Mezzo

Signal cables: Kondo Silver, Crystal Cable phono

Power cables: Kondo, Shunyata, van den Hul

system pics

Link to comment
The MQA thread made me think that all this chatter about fixing the ADC and DAC errors can, to a great extent, be accomplished as an on-the-fly process at the playback end. I expect the majority of MQA content will be post-processed digital files from the final files sent to CD production.

 

So I have two questions for Miska et al:

 

1- This de-blurring effect people are talking about - can it be implemented as a post-process filter?

 

2- In light of all the chatter of improving whatever issues are present in the input file, does this mean that upsampling might get a new think-through ie go past interpolation and further into "fixing" the original data?

 

3- The argument about tailoring the fixes to the ADC and DAC used could obviously be incorporated into settings of an on-the-fly filter - is this a possibility? (*)

 

Thx.

 

(*) iZotope in Audirvana allows you to tune some output parameters. I imagine different DACs would sound best with their tailored settings - I haven't investigated this.

 

Great topic! Was just drafting something along the same lines myself :)

John Walker - IT Executive

Headphone - SonicTransporter i9 running Roon Server > Netgear Orbi > Blue Jeans Cable Ethernet > mRendu Roon endpoint > Topping D90 > Topping A90d > Dan Clark Expanse / HiFiMan H6SE v2 / HiFiman Arya Stealth

Home Theater / Music -SonicTransporter i9 running Roon Server > Netgear Orbi > Blue Jeans Cable HDMI > Denon X3700h > Anthem Amp for front channels > Revel F208-based 5.2.4 Atmos speaker system

Link to comment

Miquel: This is similar to the question I was asking more generally on filtering in this thread http://www.computeraudiophile.com/f8-general-forum/great-filtering-does-it-matter-and-what-does-it-sound-27275/

 

I also, half joking and half seriously posted the picture below, as I think that a lot of these choices (at this level of audio refinement) become not only music dependent (different for rock than jazz or classical) but also different for what each of us is sensitive to from a hearing perspective. That is one of the things I worry about in MQA and much prefer in HQPlayer (I, not someone with different ears, a different room and a different system, get to make these fine-tuning choices).

Sound Preferences.jpg

Synology NAS>i7-6700/32GB/NVIDIA QUADRO P4000 Win10>Qobuz+Tidal>Roon>HQPlayer>DSD512> Fiber Switch>Ultrarendu (NAA)>Holo Audio May KTE DAC> Bryston SP3 pre>Levinson No. 432 amps>Magnepan (MG20.1x2, CCR and MMC2x6)

Link to comment

1- This de-blurring effect people are talking about - can it be implemented as a post-process filter?

 

Yes, it has been already implemented for a long time! It is called apodizing filter. For example poly-sinc are such, except poly-sinc-hb.

 

2- In light of all the chatter of improving whatever issues are present in the input file, does this mean that upsampling might get a new think-through ie go past interpolation and further into "fixing" the original data?

 

Yes, that is the case and is already happening depending on which filter you select...

 

3- The argument about tailoring the fixes to the ADC and DAC used could obviously be incorporated into settings of an on-the-fly filter - is this a possibility? (*)

 

ADC fixes are already in as apodizing filters. There could be some further fixes done, but it would need processing between ADC and DAW before the final mix/editing happens. MQA is also a post-processing that happens on the final master, so it cannot do those other fixes properly either.

 

I take DAC optimization further by replacing modulator in the DAC (for DSD capable DACs), MQA is still PCM-only so it is constrained on that front. I could very well add some further DAC optimizations for the DACs I have. I have a pretty good selection already and it is constantly expanding.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
Miquel: This is similar to the question I was asking more generally on filtering in this thread http://www.computeraudiophile.com/f8-general-forum/great-filtering-does-it-matter-and-what-does-it-sound-27275/

 

I also, half joking and half seriously posted the picture below, as I think that a lot of these choices (at this level of audio refinement) become not only music dependent (different for rock than jazz or classical) but also different for what each of us is sensitive to from a hearing perspective. That is one of the things I worry about in MQA and much prefer in HQPlayer (I, not someone with different ears, a different room and a different system, get to make these fine-tuning choices).

[ATTACH=CONFIG]23718[/ATTACH]

 

Agree. I would say there are "true" things to be fixed, and there also are preferences.

NUC10i7 + Roon ROCK > dCS Rossini APEX DAC + dCS Rossini Master Clock 

SME 20/3 + SME V + Dynavector XV-1s or ANUK IO Gold > vdH The Grail or Kondo KSL-SFz + ANK L3 Phono 

Audio Note Kondo Ongaku > Avantgarde Duo Mezzo

Signal cables: Kondo Silver, Crystal Cable phono

Power cables: Kondo, Shunyata, van den Hul

system pics

Link to comment
I take DAC optimization further by replacing modulator in the DAC (for DSD capable DACs), MQA is still PCM-only so it is constrained on that front. I could very well add some further DAC optimizations for the DACs I have. I have a pretty good selection already and it is constantly expanding.

Interesting... Couple of questions:

 

1- Do you have preferred settings in HQP for each DAC chip? Do you happen to have an EmmLabs Dac2x around to optimize for? :)

 

2- Can the apodizing filters be tuned to the particular ADC used? MQA's arguing it can - Could HQP have a selection of such? I imagine in all of these filters, other than the particular techniques (eg degree of polynomial or similar) there must be parameters that are chosen in these filters (steepness, etc) and I imagine this is what MQA is fine-tuning.

 

3- Do you have a rough description of filters and what you recommend - especially regarding specific DACs?

 

4- With my DAC (EmmLabs XDS1v2, identical to Dac2x) I have found that feeding it 24/192 sounds more magical (to my ears obviously) than the DSD output on HQP - letting the DAC do PCM=>DSD128 internally rather than PCC=>DSD64 (in HQP) then DSD64=>DSD128 (in the DAC). Makes sense?

NUC10i7 + Roon ROCK > dCS Rossini APEX DAC + dCS Rossini Master Clock 

SME 20/3 + SME V + Dynavector XV-1s or ANUK IO Gold > vdH The Grail or Kondo KSL-SFz + ANK L3 Phono 

Audio Note Kondo Ongaku > Avantgarde Duo Mezzo

Signal cables: Kondo Silver, Crystal Cable phono

Power cables: Kondo, Shunyata, van den Hul

system pics

Link to comment
1- Do you have preferred settings in HQP for each DAC chip?

 

Not really, other than for certain cases like Metrum Musette (no DAC chip). In most cases, filter selection is better driven by source content type (genre) and one's sonic sensitivities rather than type of DAC. For modulators there are some DAC dependencies, but I've tried to make both ASDM5 and ASDM7 such that they would be fairly generic across the board.

 

Do you happen to have an EmmLabs Dac2x around to optimize for? :)

 

No, that's too expensive... :) You can find many of my DACs listed on the HQPlayer web page under recommended hardware. Those are the ones I regularly test & measure with.

 

2- Can the apodizing filters be tuned to the particular ADC used?

 

Could, but it wouldn't make much difference.

 

MQA's arguing it can - Could HQP have a selection of such? I imagine in all of these filters, other than the particular techniques (eg degree of polynomial or similar) there must be parameters that are chosen in these filters (steepness, etc) and I imagine this is what MQA is fine-tuning.

 

MQA is also arguing that it is also generally useful for all old content even if ADC is not known. But in the end they cut word length to 17-bit which is already removing low level details of good ADCs.

 

I argue that those parameters depend more on music type and listener than the ADC. In addition lot of material ending up in RedBook format is originally recorded for example at 96 kHz sampling rate and then converted to 44.1k at mastering stage using rate conversion software.

 

3- Do you have a rough description of filters and what you recommend - especially regarding specific DACs?

 

Yes... Included in the manual. :)

 

But I prefer not to recommend anything in particular for any specific DAC unless there's a strong technical reason for doing so.

 

4- With my DAC (EmmLabs XDS1v2, identical to Dac2x) I have found that feeding it 24/192 sounds more magical (to my ears obviously) than the DSD output on HQP - letting the DAC do PCM=>DSD128 internally rather than PCC=>DSD64 (in HQP) then DSD64=>DSD128 (in the DAC). Makes sense?

 

Yes, but why don't you feed it with DSD128 straight instead of having some internal conversion?

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment

Thanks Miska for your answers. Answering your specific question below...

 

Yes, but why don't you feed it with DSD128 straight instead of having some internal conversion?

 

The USB input in the Dac2x/XDS1v2 can take DSD64 max, over DoP 1.0. I think this is a hardware limitation of the receiver chip used (xmos) since the max PCM rate is 192. There's now the option to use DoP 1.1 which I believe will let you send DSD128 over 192, but the Dac2x/XDS1v2 do not support DoP 1.1 at the moment - I tried it with Audirvana - I get a faint hiss, no music.

 

The DAC itself upsamples all PCM and DSD64 to DSD128 using their algorithm, MDAT2, and sends that to their discrete DAC.

 

It just sounds to me more "magical" using upsampled PCM to 192 than to DSD64, possibly bc there's a further internal upsampling to DSD128. The difference is very small, but the PCM just sort of grabs my attention more - whether it is more accurate or not I don't know.

 

Makes sense?

NUC10i7 + Roon ROCK > dCS Rossini APEX DAC + dCS Rossini Master Clock 

SME 20/3 + SME V + Dynavector XV-1s or ANUK IO Gold > vdH The Grail or Kondo KSL-SFz + ANK L3 Phono 

Audio Note Kondo Ongaku > Avantgarde Duo Mezzo

Signal cables: Kondo Silver, Crystal Cable phono

Power cables: Kondo, Shunyata, van den Hul

system pics

Link to comment
Your listening preferences, your result. Listening is always individual experience. It is so and it has to be so.

Right. I suppose it also stands to reason that a lot of work has gone into MDAT2 (the DSP used in the EmmLabs DACs) and they are very fine tuned to the hardware itself.

NUC10i7 + Roon ROCK > dCS Rossini APEX DAC + dCS Rossini Master Clock 

SME 20/3 + SME V + Dynavector XV-1s or ANUK IO Gold > vdH The Grail or Kondo KSL-SFz + ANK L3 Phono 

Audio Note Kondo Ongaku > Avantgarde Duo Mezzo

Signal cables: Kondo Silver, Crystal Cable phono

Power cables: Kondo, Shunyata, van den Hul

system pics

Link to comment

Very interesting questions and thoughts in this thread.

 

When using iFi micro iDSD, almost everything sounded best to me when playing as DSD256 with poly-sinc, ASDM7 and seems in-line with Miska's suggestions I have seen for this DAC.

 

When using Hugo TT, things get a bit interesting. PCM sources sound better to me when played at 384K (Poly-sinc, NS5) but not as DSD128 (Poly-sinc, ASDM7). Imaging is better, bit more airy and transparent with PCM. When playing DSD64 sources,it is harder to say but still prefer PCM 384K over DSD128. I am not sure if am the only one who feels this way and if there is a good technical explination (I would love to hear). I have read the posts by Rob Watts over at Head-fi but it is unclear if some of the benefits of HQplayer's DSD conversion would/could be lost due to how Hugo processes DSD.

 

I listen either using either He-1K or LCD-3F.

Desktop: AMD Server(AL Roon Core/ HQP NAA)  >>  Fiber >> DAVE >> (2 x HFC Trinity Helix) >> LCD-5/(KGSSHV Carbon CC + CRBN)/ Omega CAMs [Everything plugged into Sound Applications TT7]

 

Link to comment

After few days of listening my new highly customized JLsounds AK4490 DAC with HQP 3.13b2 on Linux Mint + Miska patches:

 

DAC settings: SDM Pack(none)/Buffer(50)/DAC bits(32)

Listening preference in order of priority:

 

[1] DSD64>DSD256 - closed-form/DSD256+fs (or ASDM7)/11.2

[2] PCM16/44 native - none/none/44,1

[3] PCM16/44>DSD128 - minringFIR/ASDM5/5.6

 

Upsampling DSD64 file to DSD256 is definitely the winner, the beauty of DSD signature with PCM like attacks and speed. For an unknown magic combination native 16/44 sounds now wonderful with my new dac[/url], really top class so I must rank it as #2 (just because DSD has more carry more informations) ... Then PCM>DSD has it's own space with some 16/44 translated to DSD128 ... but in general I don't like the result due to the fact sounds lose focus and the executors become too big on stage.

 

Have a nice day, Massimiliano

Link to comment
After few days of listening my new highly customized JLsounds AK4490 DAC with HQP 3.13b2 on Linux Mint + Miska patches:

 

DAC settings: SDM Pack(none)/Buffer(50)/DAC bits(32)

Listening preference in order of priority:

 

[1] DSD64>DSD256 - closed-form/DSD256+fs (or ASDM7)/11.2

[2] PCM16/44 native - none/none/44,1

[3] PCM16/44>DSD128 - minringFIR/ASDM5/5.6

 

Upsampling DSD64 file to DSD256 is definitely the winner, the beauty of DSD signature with PCM like attacks and speed. For an unknown magic combination native 16/44 sounds now wonderful with my new dac[/url], really top class so I must rank it as #2 (just because DSD has more carry more informations) ... Then PCM>DSD has it's own space with some 16/44 translated to DSD128 ... but in general I don't like the result due to the fact sounds lose focus and the executors become too big on stage.

 

Have a nice day, Massimiliano

Awesome DAC pics... Can you provide details?

NUC10i7 + Roon ROCK > dCS Rossini APEX DAC + dCS Rossini Master Clock 

SME 20/3 + SME V + Dynavector XV-1s or ANUK IO Gold > vdH The Grail or Kondo KSL-SFz + ANK L3 Phono 

Audio Note Kondo Ongaku > Avantgarde Duo Mezzo

Signal cables: Kondo Silver, Crystal Cable phono

Power cables: Kondo, Shunyata, van den Hul

system pics

Link to comment

In the User Manual the description for poly-sinc-short is:

Otherwise similar as poly-sinc, but shorter pre- and post-echos at

the expense of filtering quality (not as sharp roll-off and reduced

stop-band attenuation)

 

Would this make poly-sinc-short more suited to only higher source PCM sample rates, eg: 4x and above? and everything below this, for example 44.1k more suited to poly-sinc & poly-sinc-mp?

 

How does the roll-off curve differ for poly-sinc-ext vs the others? edit: never mind - it is described in the current user manual. I was looking at an older version of the manual online.

Link to comment
It just sounds to me more "magical" using upsampled PCM to 192 than to DSD64, possibly bc there's a further internal upsampling to DSD128. The difference is very small, but the PCM just sort of grabs my attention more - whether it is more accurate or not I don't know.

 

Makes sense?

 

It largely depends on how DSD64 -> DSD128 is implemented inside the DAC versus how PCM192 -> DSD128 is implemented.

 

One test you could also perform (with hqplayer 3.13b3+) to compare is to play back some DSD64 content converted to PCM by HQPlayer. For this test, select PCM output mode, PCM filter "none" and dither "TPDF", then in DSDIFF/DSF settings dialog noise filter "standard" and then you can try various different conversion algorithms to see how it compares to how the DAC plays DSD64 by itself. This will produce 176.4k PCM output.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment

Would this make poly-sinc-short more suited to only higher source PCM sample rates, eg: 4x and above? and everything below this, for example 44.1k more suited to poly-sinc & poly-sinc-mp?

 

Both are equally suitable... I need to actually remove that reduced stop-band part which is not the case anymore in recent versions, only difference is steepness of the roll-off slope.

 

I use the lp/mp short variants myself. The new -mqa version is more extreme and combination of -ext with even slower roll-off than "short" and tuned to roll-off in a way that it cuts into HF noise produced by MQA encoder.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
Both are equally suitable... I need to actually remove that reduced stop-band part which is not the case anymore in recent versions, only difference is steepness of the roll-off slope.

 

I use the lp/mp short variants myself. The new -mqa version is more extreme and combination of -ext with even slower roll-off than "short" and tuned to roll-off in a way that it cuts into HF noise produced by MQA encoder.

I would actually like a setup like the 'custom' you have in Audirvana where I can say for each input rate what algo/output I want. This is obviously a fairly major rework of the GUI, understood.

NUC10i7 + Roon ROCK > dCS Rossini APEX DAC + dCS Rossini Master Clock 

SME 20/3 + SME V + Dynavector XV-1s or ANUK IO Gold > vdH The Grail or Kondo KSL-SFz + ANK L3 Phono 

Audio Note Kondo Ongaku > Avantgarde Duo Mezzo

Signal cables: Kondo Silver, Crystal Cable phono

Power cables: Kondo, Shunyata, van den Hul

system pics

Link to comment

...

 

It just sounds to me more "magical" using upsampled PCM to 192 than to DSD64, possibly bc there's a further internal upsampling to DSD128. The difference is very small, but the PCM just sort of grabs my attention more - whether it is more accurate or not I don't know.

 

Makes sense?

 

It largely depends on how DSD64 -> DSD128 is implemented inside the DAC versus how PCM192 -> DSD128 is implemented.

 

One test you could also perform (with hqplayer 3.13b3+) to compare is to play back some DSD64 content converted to PCM by HQPlayer. For this test, select PCM output mode, PCM filter "none" and dither "TPDF", then in DSDIFF/DSF settings dialog noise filter "standard" and then you can try various different conversion algorithms to see how it compares to how the DAC plays DSD64 by itself. This will produce 176.4k PCM output.

 

Miska that's a possibility as well, but more than likely, it's the quality of media production (recording quality, mastering etc.) as the more plausible explanation for this.

 

DSD and PCM are simply transfer protocols. As such, it's irrelevant which is used for base media production.

 

BTW, my settings are poly-sinc mp and RPDF.

Link to comment
It largely depends on how DSD64 -> DSD128 is implemented inside the DAC versus how PCM192 -> DSD128 is implemented.

 

One test you could also perform (with hqplayer 3.13b3+) to compare is to play back some DSD64 content converted to PCM by HQPlayer. For this test, select PCM output mode, PCM filter "none" and dither "TPDF", then in DSDIFF/DSF settings dialog noise filter "standard" and then you can try various different conversion algorithms to see how it compares to how the DAC plays DSD64 by itself. This will produce 176.4k PCM output.

 

I am a registered user, but how can I download hqplayer 3.13b3+ ?

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...