HQPlayer settings for specific DACs

[Miska]

On 1/22/2022 at 5:50 PM, ted_b said:

In HQplayer the filter settings (for PCM, whether upsampling to PCM or to DSD) and the modulator settings (for DSD) are often quite personal.  Why?  Like any other component in your signal path stack, the decisions you make for sound quality have everything to do with your own:

* room acoustics

* speaker (or headphone) capabilities

* listening preferences and hearing issues

* music styles and recording qualities

* (and in the case of HQPlayer) cpu/server power and demands

 

It is, therefore, somewhat superfluous to take others' recommendations when you know nothing of their systems or preferences.  Listen for yourself and make filter and modulator decisions, just as you would sample rate decisions, based on the equipment, room and listening preferences you are "stuck with".

 

OTOH, setting things like dac bit depth can be cataloged by dac.  In the case of the Holo May, for example, setting bit depth to 20 bits is the May's best setting for linearity and least noise. 

 

For modulator choice, one big determining factor is also DAC architecture, but mostly in terms of choosing between 5th and 7th order modulators. And as special case the M-modulators.

 

DAC Bits is purely choice based on DAC hardware.

 

[Miska]

4 hours ago, TJHUB said:

As Ted mentioned above, I set my DAC Bit to "Default" and easily got the best sound.  This sets the DAC to 32 bit.  Definitely start there.  I've tried many other settings, and none were nearly as good.

 

Which DAC is it?

 

[Miska]

17 minutes ago, ted_b said:

Terry has an Eastern Electric Minimax Junior dac

 

Ahh, OK, it is ESS. For those, 32-bit (default) output is correct.

 

[Miska]

24 minutes ago, sbenyo said:

After trying a few options, I am able to use either ASDM5ec or ASDM7ec 512+fs at DSD256. Not sure what is the difference.

 

Those 512x+ versions are AMSDM, which makes those completely different from the others. Only recommended for DSD rates of 512, 1024 and so on. Although allowed for 256 rates too.

 

If you can use ASDM5EC, I'd recommend using it at DSD256. Will give you very good performance.

 

24 minutes ago, sbenyo said:

I also trying to decide between ext3 and xtr-lp. Still not sure which one is better.

 

There's no clear better for choosing between these, which ever sounds better to you. Technically both are very good, but different.

 

I would recommend to pay attention to the "apodizing count" in the GUI. It is running analysis on the source as the playback proceeds. If the counter increases to values over 10, I would strongly suggest choosing one of the apodizing filters (refer to the manual table).

 

[Miska]

1 hour ago, sbenyo said:

Update. Waiting longer it does get to high number with ext3 which currently sounds better to me.

 

Filter choice doesn't affect Apod counter value, as it result of analysis run on the input data before any DSP is applied by HQPlayer. Analysis proceeds together with the playback, so you know final figure at end of the track.

 

[Miska]

54 minutes ago, creativepart said:

I'm pretty sure I saw Miska say to leave "Dac Bits" blank.

 

I'm not exactly sure where I saw that - but that's my memory.

 

Depends on a DAC and how you interface to the DAC.

 

For Holo Audio DACs, 20 bit is correct setting.

 

[Miska]

5 minutes ago, FooFighter said:

Miska, any recommendations for filter and eq settings withing HQPlayer to make Susvaras more full bodied and have some mid bass thump on Spring 3 KTE attached to Ferrum OOR for Rock and Pop genre?

 

Have you tried the Oratory EQ settings from here?

https://www.dropbox.com/s/77xp4wrohjgwqzq/Hifiman Susvara.pdf?dl=0

 

You can use REW to create EQ .txt file from those. Or alternatively just enter those values straight to HQPlayer. You can use the Plot function to check that the response matches with the one on on PDF so you know you there are no errors in the settings.

 

Sounds like poly-sinc-short-mp(-2s) could be a suitable filter for your case.

 

[Miska]

3 hours ago, FooFighter said:

I am now wondering if I can reach the same thing by using an exported WAV-file out of the REW application and apply that one to the HQ Player convolution engine.

 

It is better to stick with the parametric EQ, either straight or through .txt export from REW instead of convolution. You save the unnecessary step of converting those parametric EQs to convolution filter and also CPU time processing the correction at HQPlayer side.

 

3 hours ago, FooFighter said:

As far as I read that seems to allow me creating different PEQs-convolution profiles for different headphones and I can seamlessly switch between these settings whereas entering a pipeline setup will interrupt the playback.

Correct?

 

Yes, you can create various different matrix profiles and switch those on the fly for example using HQPlayer Client. No need to even stop the playback.

 

I have various, for different headphones plus the loudspeaker system plus multichannel mix-down for loudspeakers and headphones.

 

3 hours ago, FooFighter said:

First questions I have are about settings I should select for exporting as WAV if I don't know the bit rates and sample rates I am playing in advance?

Or should I just select 24bit and check all boxes?

Do you happen to know if I am supposed to select Manual or Auto as Control-type as I don't understand what the REW-app is supposed to do automatically for a manually specified PEQ

 

 

If you enter it this way, just export is as generic EQ .txt file instead of convolution filter.

 

3 hours ago, FooFighter said:

Next question I have is about the parallel usage of the pipeline and the convolution engine.

If I just want to apply the PEQ settings I don't need to set both - as for the described better comfort of changing saved convolution profiles I can leave the pipeline settings completely disabled, correct?

 

If you use matrix pipeline, you need to disable convolution engine. These two are mutually exclusive. Matrix can do everything convolution engine can do and more. You can load parametric EQ .txt files or convolution filter WAVs there, in addition to the specs you can enter straight as text there.

 

 

[Miska]

16 hours ago, askat1988 said:

@Miska Can't we just use the .txt files from Github?  Here is an example: https://github.com/jaakkopasanen/AutoEq/tree/master/results/oratory1990/harman_over-ear_2018/RAAL-requisite SR1a

 

Which one is the correct one, fixed, graphic, parametric?

 

Parametric .txt files.

 

But note that these are auto-generated from external data. So for example compare the settings to the ones listed by Oratory if the correction you are looking for is sourced from there. If your headphones have more than one setting available in AutoEq, it makes also sense to try different ones to find one that best suits you personally.

 

[Miska]

9 hours ago, ccc888 said:

How many bits should I set for my Singxer SDA6, which has a AK4499 Chip?

I upsample everything to 768/705.6KHz PCM under NOS mode.

 

32 bits.

 

9 hours ago, ccc888 said:

And what should I set if I upsample to DSD in the future?

 

AK4499 performs best at DSD512.

 

9 hours ago, ccc888 said:

Also Could you provide more detail on why Saber DAC bits setting is different?

 

Different compared to...?

 

9 hours ago, ccc888 said:

What is the principle of choosing DAC bits for different DACs?

 

When DAC performs DSP for the PCM input data (any delta-sigma DAC such as AK4499), it should be set to maximum resolution the DSP can utilize. Taking into account any word length limitations on the way there. Such as S/PDIF or AES/EBU which are limited to 24-bit.

 

When it is R2R DAC, you should be utilizing linear region of the DAC based on linearity sweep measurement. If the sweep indicates the linearity is somewhere between two word lengths, pick the lower value. It is better to use one bit less than the linear region, rather than one bit too much.

 

 

[Miska]

8 hours ago, Mark Labbett said:

what would this be for the Ares II?

 

16-bit is the correct setting for Ares II.

 

[Miska]

4 hours ago, Mark Labbett said:

@Miska

i have a problem specific with HQPlayer in that it will only allow me to select DoP and as such i am unable to do DSD1024 (DSD512 is fine with DoP), i had this problem once before and i fixed it by uninstalling HQPlayer and re-installing it, then a few weeks later my computer crashed and i loaded a system image but the problem has returned and nothing i do has fixed it.

As i stated earlier it is only on HQPlayer that i cannot do native DSD, if i try HQPlayer just shuts down 

Foobar and Jriver all do it without issue so i believe this is a bug in HQPlayer, is there a log that i can send to you to see if you can identfy the problem. Thanks in advance.

 

This sounds like you are on Windows? Usually this problem appears with certain drivers (Thesycon) when there's more than one application using the audio device at the same time. For example if Roon is running simultaneously and has the driver enabled as output zone (Roon Settings -> Audio).

 

[Miska]

5 hours ago, Mark Labbett said:

Thanks for the reply but this was one of the first things i checked, as i said previously native DSD works with Foobar and Jriver it is only HQPlayer that will only allow playing via DoP, is there a crash log generated when HQPlayer crashes? because that may hold the answer

 

If HQPlayer crashes, then it is likely the ASIO driver that crashes for some reason.

 

You can enable HQPlayer log file and try to catch the last items that got written before the crash.

 

I cannot comment about Foobar or Jriver, I have never used such.

 

[Miska]

1 hour ago, Mark Labbett said:

i have given the links to the crash dump and HQPlayer log in previous post i am hoping that i don't have to re-install windows fresh to fix this problem so any help you can provide will be most appreciated!

 

You seem to have "AltDSD" enabled in settings. Meaning "pdm_div" in settings.xml set to "1" instead of "8" (default). This will certainly crash the application if it's not required by the driver (none of the drivers currently on market). I have removed that option from the GUI.

 

You can either manually edit the settings file (use the backup / restore option). Or alternatively reset the configuration through the Help-menu option and start from fresh.

 

[Miska]

14 hours ago, Reg19 said:

 

     would the SQ from both be very similar to each other? In other words, does upsampling to DSD256 (or DSD512 if usable) reduce the differences in SQ between generally well-measuring DACs that play well with HQP even if, subjectively without using HQP, one seems better than the other?

 

Assuming here the ADI-2 is AKM chip one. Fairly similar yes, but some differences remain due to different kind of D/A conversion stage implementation and following analog filter stages. So yes the differences become smaller.

 

[Miska]

3 hours ago, Soundfx222 said:

I am a newbie with HQ player commercial version and have a question regarding the settings on my DAC which is the Pontus II. If I am using HQ player for my up sampling should I enable over sampling from the Pontus menu or NOS so that HQ player does the job without interference from the DAC trying to performance over sample as well? Would there be a conflict with both set to over sample? Thank you.

 

Audio path = HQ player → PC (USB to ) →MiniDSP SHD (spidif out to) →Pontus → Amp

 

First you should remove the MiniDSP thing from the path, because it is limiting sampling rate. Just connect Pontus directly using USB.

 

Set Pontus to NOS mode and HQPlayer to output 705.6/768k rate. Set DAC Bits to 16 for starters and use LNS15 dither/noise-shaper.

 

You can also try with DSD256 for comparison.

 

[Miska]

3 hours ago, Ozzie said:

What DAC bits and Filter settings would you recommend for a Yggdrasil Less is More (its a 16-bit TI DAC)? I have it set to 24-bits and 192k Rate PCM with poly-sinc-gauss-long. 

 

I should also mention that I am using Allo DigiOne HAT on Pi4 BNC output to BNC input

 

At most 16-bit. I recommend TPDF or Gauss1 dither, but you can also try NS9 noise-shaper.

 

[Miska]

1 hour ago, Ozzie said:

Thanks! NS9 works pretty well. 

 

Does reducing the DAC bit output not affect the quality if I am also adding a -3db attenuation? Or am I not thinking about this right? I thought that the volume bits would reside beyond the 16-bits and so a 20-bit would be better, no? 

 

The DAC chip in question can only use 16 bits. Bits beyond those are just thrown away at the DAC chip which generates distortion.

 

Using correct number of output bits improves quality because then you don't get distortion due to either truncation and/or non-linearity. So using less bits gives better quality. In addition, when your output rate is high enough and you use noise-shaping, there's no loss in resolution, but there can be increase in resolution, without loss in audio band dynamic range.

 

[Miska]

56 minutes ago, andrewd01 said:

Do you have any recommended settings for DSD upscaling to a AK4490EQ DAC chip?  I am using the embedded version of HQP.

 

Both ASDM7EC(v2) and ASDM5EC(v2) modulators work fine for running it at DSD256 rate. I recommend the former one. With filter of your choice.

 

If possible, try to ensure the chip is in DSD Direct mode.

 

[Miska]

38 minutes ago, andrewd01 said:

Thanks, the DAC is an accuphase DAC50.  The specs for DSD shown below.

I understand that 11.2896 MHz is DSD256.  Does the ‘ASIO only’ comment for the high rate mean that rate is not ‘DSD direct’?

 

https://www.accuphase.com/model/dac-50.html

DSD/2.8224MHz,5.6448MHz,11.2896MHz(1bit)
                    *11.2896MHz(ASIO only)

 

No, it means that the rate only works on Windows through their ASIO driver. I don't know how the DAC chip is configured for DSD on the DAC50 card. You could try to ask Accuphase if they can tell whether they use DSD Direct mode or not. Since they don't need the on-chip volume control, I find it likely that the chip is indeed used in DSD Direct mode.

 

[Miska]

1 hour ago, sbilotta said:

Hi, I'm evaluating HQP within Roon and using a MSB Diamond DAC V.

I upsample to DSD256 in HQP and perform convolution filters in Roon (see pic below).

 

The issue I'm finding is that PCM in multiples of 48K don't play in HQP, BUT... 48K itself does play...

I've tried several bit rate settings, but to no avail.

Below is what I have in settings.

Can you please advise what should be the correct setting?

 

 

Thank you!!!

 

Reason is that you have poly-sinc-gauss-long filter selected for 1x rates, which can convert between rate families. While minringFIR-mp filter selected for Nx rates cannot. Please refer to the filter table in the manual.

 

 

P.S. I would move the convolution from Roon to HQPlayer. And leave Roon as bit-perfect transport.

 

[Miska]

5 hours ago, sbilotta said:

I had Roon handle the convolution filters because my NUC (NUC10i7FNHN) couldn't handle HQP doing both upsampling and convolution on DSD128 and 256, while separating the load yes.

 

OK, I see. This is better than doing what you describe below. Best option would be to have enough CPU power so that you don't need to resort to such due to lack of CPU power.

 

5 hours ago, sbilotta said:

However, I have now found that I can use the DirectSDM setting (in DSD Source Settings) and let Roon upsample only DSD64 and 128 to DSD256, allowing HQP to always receive DSD256 (also from PCM upsampling) and apply convolution filters.

 

This is bad idea. If you have DirectSDM enabled, HQPlayer won't be doing any DSP, convolution or anything else than speaker delays for DSD inputs. Essentially in this case HQPlayer is doing nothing.

 

5 hours ago, sbilotta said:

Could you please advise the most appropriate filters in the SDM Defaults and DSD Source Settings for my DAC?

 

If you have DirectSDM enabled, none of those have any effect. But I would recommend to get started with FIR2 + XFi.

 

[Miska]

2 hours ago, all300b said:

1.  Is there an easy way to confirm which convolution filter is actually running during playback- other than clicking into the sub-menu of the client? Especially when using Roon to HQplayer it would be nice for the regular player to indicate whether and which convolution filter is running.

 

If you mean matrix profiles, no, not from the server window. You can see if the simple convolution engine is active or not. But many of the runtime options are not available from the server window (active filter, dither/modulator, etc).

 

2 hours ago, all300b said:

2.  Any special filter recs for a topping D90 DAC (AKM4499) for someone who - no matter how many times I try Gauss, Sinc, and other new filters - keeps coming back to poly-sinc-short-minimum phase for rebook (upsampled to 384).

Thanks!

 

AK4499 works nicely at DSD512. So run it with ASDM7EC(v2) or ASDM5EC(v2) and you are good. Filter choice is up to personal preferences combined with the source material (genre mostly). poly-sinc-short-mp is just fine! IMO, at it's best with prog rock.

 

[Miska]

1 hour ago, Schafheide said:

To quote the manual

However, using “[source]” usually leads to sub-optimal result with either format since
only very few DACs have separate true PCM (R2R) and SDM conversion sections
inside. In most cases only either of the options is optimal for the DAC.

Is it ok to select "source" for the Holo May DAC?

 

Yes, one of the very few. Although it may still not be the most optimal. But at least it has separate converter sections for both, so you are really going through different paths.

 

[Miska]

7 hours ago, Schafheide said:

So, what should I select in the following scenarios?

1. PCM file

2. DSD file

 

or would "source" yield the same results?

Either way, is it best to stick to PCM - PCM for O/S and DSD - DSD ?

 

I'm sticking to running my Holo Springs at DSD256 (Spring 2) or DSD512 (Spring 3). So I have output mode always set to SDM.

 

Especially if you use "[source]" setting, remember to set PCM gain compensation to -6 dB in HQPlayer settings. This is also necessary if you want to compare PCM vs DSD output modes.