HQ Player

[El Guapo]

1 minute ago, dericchan1 said:

plot the graphs for channel 1 and channel 2 now

Looks like right channel has +2dB @ 1200Hz? Suggest add preamp parameter to the first line of both L, R PEQ txt:

Preamp: -2 dB

 

[Miska]

4 minutes ago, dericchan1 said:

Thanks @El Guapo

 

with your assistance I was able to plot the graphs for channel 1 and channel 2 now

 

 

 

Here, Channel 2 has 2 dB boost, so you should set -2 dB mix gain for both pipelines to compensate for that narrow boost.

 

[dericchan1]

3 minutes ago, El Guapo said:

Looks like right channel has +2dB @ 1200Hz? Suggest add preamp parameter to the first line of both L, R PEQ txt:

Preamp: -2 dB

 

Thanks for catching that, it was meant to be -2 but missing a 2 in the txt file. Do I need to delete the process in hqplayer then re-link?

[Miska]

5 minutes ago, dericchan1 said:

Thanks for catching that, it was meant to be -2 but missing a 2 in the txt file. Do I need to delete the process in hqplayer then re-link?

 

As long as you keep the same file name you can just edit the file. It gets re-read when you start playback next time. HQPlayer keeps reference to the file itself and doesn't store it's content by itself.

 

[dericchan1]

16 minutes ago, Miska said:

 

As long as you keep the same file name you can just edit the file. It gets re-read when you start playback next time. HQPlayer keeps reference to the file itself and doesn't store it's content by itself.

 

Thanks Miska, also what does load, save, delete at the top for? It does not appear to do anything when I clicked on those?

[Miska]

9 minutes ago, dericchan1 said:

Thanks Miska, also what does load, save, delete at the top for? It does not appear to do anything when I clicked on those?

 

On the left side you have list of profiles and you can enter a new name there. If you type a name and hit Save, new matrix profile is saved. When select something from the list and hit Load, it will the loaded to the matrix setup. If you hit Delete, the selected profile is deleted.

 

When you leave the dialog by pressing OK, the currently active/loaded profile is saved as default that is loaded when HQPlayer is started.

 

You can switch between the profiles on the fly from control applications such as HQPlayer Client.

 

I have separate profiles for different headphones I'm using and in addition things like loudspeaker system correction. And then variants with 5.1 channel to stereo mixdown. Although you can listen to stereo with 5.1 mixdown profile all the time just fine, there's just couple of dB level loss for stereo sources due to mix gains.

 

 

I also have a setup for Merging Hapi where first two output channels from Hapi go to loudspeaker system and second two go to Ferrum Oor with headphone corrections:

So above matrix duplicates stereo source to output channels 1+2 and 3+4. I could also do something like main speakers on channels 1+2 and subwoofer low-pass and correction for outputs 3+4, etc. Hapi has 8 output channels available.

 

[StreamFidelity]

@Miska 

I am wavering between two modulators for DSD512:

 

AMSDM7EC 512+fs
Special adaptive seventh order "pseudo-multi-bit" modulator with extended compensation for rates >= 20.48 MHz

or 

ASDM7ECv2
Second generation of ASDM7EC with minor improvements.

 

With fanless PC (i9-12900K) I can use both. AMSDM7EC 512+fs has a slightly lower load (maybe 2% less) than ASDM7ECv2.

 

Where are the qualitative differences? What do you recommend?

[Miska]

9 hours ago, StreamFidelity said:

@Miska 

I am wavering between two modulators for DSD512:

 

AMSDM7EC 512+fs
Special adaptive seventh order "pseudo-multi-bit" modulator with extended compensation for rates >= 20.48 MHz

or 

ASDM7ECv2
Second generation of ASDM7EC with minor improvements.

 

With fanless PC (i9-12900K) I can use both. AMSDM7EC 512+fs has a slightly lower load (maybe 2% less) than ASDM7ECv2.

 

Where are the qualitative differences? What do you recommend?

 

It is hard to say which one would be better. But it largely also depends on your DAC. AMSDM puts more demands on the analog filter than regular modulators. But it has very unique way to solve certain D/A conversion problems.

 

For example for something like Holo Audio, I would recommend AMSDM7EC for DSD512/DSD1024, it can comfortably deal with such.

While for something like ESS DAC chips, I would recommend to stick to ASDM5ECv2. Although for ESS DAC chips too it depeds a lot what kind of post DAC chip analog filter stage there is. It is entirely possible to make ESS DAC perform very nicely with something like AMSDM7EC as well, devil is in the details... So the overall guidance  regarding DAC chips applies to "typical implementation".

 

[dericchan1]

Hi Miska, so I imported the peq txt file for my left and right speaker channel and it works fine for converting all pcm source files to dsd256 with 7ECV2.

 

however converting from dsd source to dsd256 with  7ECV2 now stutters. Is matrix pipeline configuration being handled by the gpu if cuda offload is enabled? I wonder if it has to do with My gtx980ti not sufficient to process the pipeline function?

 

thanks

 

Deric

[Miska]

54 minutes ago, dericchan1 said:

however converting from dsd source to dsd256 with  7ECV2 now stutters. Is matrix pipeline configuration being handled by the gpu if cuda offload is enabled? I wonder if it has to do with My gtx980ti not sufficient to process the pipeline function?

 

At the moment, parametric EQ is running on CPU. So it doesn't depend on the GPU. I would also suggest to experiment with Multicore DSP set to either checked or grayed (force-all/auto).

 

[dericchan1]

2 minutes ago, Miska said:

 

At the moment, parametric EQ is running on CPU. So it doesn't depend on the GPU. I would also suggest to experiment with Multicore DSP set to either checked or grayed (force-all/auto).

 

Thanks Miska, I suppose then it’s my i7 6700k not up to par to run peq then. I will try the multi core dsp options tonight 

 

Deric

[dericchan1]

3 hours ago, Miska said:

 

At the moment, parametric EQ is running on CPU. So it doesn't depend on the GPU. I would also suggest to experiment with Multicore DSP set to either checked or grayed (force-all/auto).

 

Hi Miska, I now changed multi core dsp from grayed to now checked. I can now play dsd64 source files upsampled to dsd256 7ECV2 but dsd128 source files stuttered, that’s an improvement!! Cheers

 

Deric

[wanta911]

I've been using HQ Player to it's full up-sampling capability with headphone EQ via the convolution pipeline for a while now 👍

 

I want to try it in my speaker setup with the iFi Zen Stream as a NAA to compare against the Volumio based player but will only up-sample to 96khz or keep as 44.1khz as I use digital room correction via MiniDSP & Dirac and the MiniDSP unit resamples everything to 96khz anyway.

 

Reading the documentation, I'm guessing that using the TPDF noise shaper is the best bet for PCM at these sample rates? I'm also guessing that any poly-sinc filter will be fine?

 

Any feedback or experience would be welcome.

[Miska]

10 hours ago, wanta911 said:

I want to try it in my speaker setup with the iFi Zen Stream as a NAA to compare against the Volumio based player but will only up-sample to 96khz or keep as 44.1khz as I use digital room correction via MiniDSP & Dirac and the MiniDSP unit resamples everything to 96khz anyway.

 

Using 96k would be best option when using such device.

 

But optimal results would be to move all the room correction to HQPlayer.

 

10 hours ago, wanta911 said:

Reading the documentation, I'm guessing that using the TPDF noise shaper is the best bet for PCM at these sample rates? I'm also guessing that any poly-sinc filter will be fine?

 

TPDF or Gauss1 are fine. Most poly-sinc filters work for this case, but you can refer to the manual for details regarding each filter.

 

[Kalpesh]

On 7/18/2022 at 1:24 AM, Miska said:

 

It is hard to say which one would be better. But it largely also depends on your DAC. AMSDM puts more demands on the analog filter than regular modulators. But it has very unique way to solve certain D/A conversion problems.

 

For example for something like Holo Audio, I would recommend AMSDM7EC for DSD512/DSD1024, it can comfortably deal with such.

While for something like ESS DAC chips, I would recommend to stick to ASDM5ECv2. Although for ESS DAC chips too it depeds a lot what kind of post DAC chip analog filter stage there is. It is entirely possible to make ESS DAC perform very nicely with something like AMSDM7EC as well, devil is in the details... So the overall guidance  regarding DAC chips applies to "typical implementation".

 

with Holo Audio you recommend ASDM7EC over the V2 version ?

and which SDM integrator ? SDM conversion ?

[Miska]

1 hour ago, Kalpesh said:

with Holo Audio you recommend ASDM7EC over the V2 version ?

and which SDM integrator ? SDM conversion ?

 

ASDM7ECv2 or ASDM7EC. Former is improved version over the latter, but takes some more CPU time.

 

But there's no AMSDM7ECv2, just AMSDM7EC. Technically it would be possible to make v2 also of the "multi-bit version".

 

[musicjunkie917]

@Miska

 

With is better for the May, ASDM7ECv2 or  AMSDM7EC?

[Miska]

1 minute ago, musicjunkie917 said:

With is better for the May, ASDM7ECv2 or  AMSDM7EC?

 

Not related to the DAC, but ASDM7ECv2 is technically slightly better than ASDM7EC. The difference is very small though.

 

From DAC's perspective, both are the same.

 

[Account Closed]

Jussi,

 

An observation and a question. I used to use rate family conversion all the time but when I had to disable it when I wanted to try gauss xla ( my M1 could not handle 44.1--> 48K conversion at EC7/256) I noticed that the sound quality seemed slightly better with it off. I have a Spring 2 so I could easily do either 44.1 or 48K family to the DAC. I may be imagining it, but the sound did seem slightly more relaxed and natural with it off. When I went back to gauss long ( could not tell any difference from xla) I decided to leave adaptive fully checked. Now the cpu usage is at about 40% and I still think it sounds better that way.

 

Is this to be expected? I know that 48K --> 44.1 is necessary for many DACs but is there any anticipated benefit for going 44.1K--> 48K family for a DAC like the Spring2 that can accept either rate family?

 

Thanks

[EMINENT]

Is it bad to use HQP volume through Roon since NOS mode removes volume control on my new R26? I have DAC bits set to 20 as I have read for R2R.

[kelvinwsy]

May I assume the R26 is the new Gustard R36 NOS DAC?

Any SQ Impressions to share vs what DAC you may hv been using?

Thx

 

[kelvinwsy]

Sorry typo Gustard R26

[Miska]

6 hours ago, bobflood said:

Is this to be expected? I know that 48K --> 44.1 is necessary for many DACs but is there any anticipated benefit for going 44.1K--> 48K family for a DAC like the Spring2 that can accept either rate family?

 

Luckily 48k -> 44.1k family is quite a bit lighter than 44.1k -> 48k family.

 

There is slight advantage of going from 44.1k to 48k family as it allows somewhat higher rates without quality trade-offs. At the cost of burning more CPU cycles (or well, more cache/RAM cycles to be precise). More CPU cache you have, better it works.

 

[Miska]

3 hours ago, EMINENT said:

Is it bad to use HQP volume through Roon since NOS mode removes volume control on my new R26? I have DAC bits set to 20 as I have read for R2R.

 

No, technically just better. Only down side is that software volume control is inherently more susceptible to accidents (more dangerous) than hardware ones.

 

[AudioDoctor]

I think we have to stop the thread here until we can get to DSD rates higher than 1024...