HQ Player

[jimdukey]

When upsampling everything to DSD 128,

what is the advantage of having a different Filter for 1x and nx?

I like the Polysincext-2 Filter a lot, esp after a shoot-out of sorts on "Speak Like a Child" 24/192.

It just seemed to snap into place and sounded perfect when I went to PSE-2!

Is having the same 1x and nx Filter ok?

I really appreciate the readout of Mode ,Filter, Rate, etc in the left corner area!

[Ipoci]

I recently changed changed my computer as a source with the following components:

- mini-ITX mobo Gigabyte GA-Q87 equipped with i5-4590 processor in a Silentmax passive chassis (very nice)

- NAA BeagleBone Black connected to DSC2.5.2 DAC

- NAS Drobo 5N

In terms of performance,  the unit can upsample 44kHz  to minringFIR, ASDM7, DSD256 ... unfortunately ClosedForm or Closed-Form-16M don't work but I' satisfied.

 

In terms of LAN setup, currently I've a plain setup using a single port:

- Pc, NAS and NAA are all IPv4 into the same subnet of the default Gateway 192.168.1.x/24 with enabled DHCP (so all dynamic)

 

Do you have any experience with dual LAN setup? I mean a LAN dedicated to NAA and the second one for NAS ... 

 

Have a nice day. Massimiliano

 

 

 

 

 

[Miska]

1 hour ago, jimdukey said:

When upsampling everything to DSD 128,

what is the advantage of having a different Filter for 1x and nx?

I like the Polysincext-2 Filter a lot, esp after a shoot-out of sorts on "Speak Like a Child" 24/192.

It just seemed to snap into place and sounded perfect when I went to PSE-2!

Is having the same 1x and nx Filter ok?

I really appreciate the readout of Mode ,Filter, Rate, etc in the left corner area!

 

Hires productions are usually better, and they also hit the filter less. So this is a way to have a different filter for hires and RedBook. But it is also totally fine to have same filter for both (v3 way). On my desktop headphone system I have ext2 filter for both. On the speaker system I have poly-sinc-mqa-lp for Nx and ext2 for 1x.

 

[jimdukey]

Thanks, Miska, I'll try that!

[koldby]

4 hours ago, Le Concombre Masqué said:

why not use your convolution filters in hqp? and you should go the extra mile and use RePhase after REW

But I am going to use the convolution engine in HQ, but I have a pre-EQ BEFORE the convolver file has been made by Audiolense, so I need to have the same EQ before the convolution in HQ. HOW DO I DO THAT?

[koldby]

4 hours ago, Le Concombre Masqué said:

why not use your convolution filters in hqp? and you should go the extra mile and use RePhase after REW

Why should I use RePhase after REW. I only use REW to make the pre-EQ. Audiolense is doing the work after that no need for RePhase then!

[Miska]

17 minutes ago, koldby said:

But I am going to use the convolution engine in HQ, but I have a pre-EQ BEFORE the convolver file has been made by Audiolense, so I need to have the same EQ before the convolution in HQ. HOW DO I DO THAT?

 

Why do you have two? Anyway, you can collapse both into one...

 

I'll add more flexibility to the matrix config once I get over some of the new GUI hoopla...

 

[robi20064]

7 hours ago, Superdad said:

 

Thanks for your replies.

 

Since:

a) 4.10 Embedded is where you say you added auto-discovery by Client;

and

b) NAA discovery is currently broken in 4.10 (and you are right, I can’t get off Config screen in 4.10)

then

c) It makes sense that Client won’t see HQP Embedded server.

 

As a test I can run a long USB cable from the HQPe computer direct to the DAC, skipping the NAA NUC for the moment (though SQ-wise it won’t take me where I want to go).

 

Alternatively, I can boot HQPe 4.9 (which does see NAA) and you can tell me what to type in the Client drop down so that it can connect to the HQPe server.

 

That will still leave me clueless about getting music tracks into a library or directly loaded.

 

In HQPe, what is the “Input” screen for?  The drop down show all sorts of things like CD Drive, Removable Media, and then a bunch of USB listings with sample rates.  But selecting any of those does not take—it just reverts to Playlist.

 

All very confusing.  Is there a manual for HQP Embedded?

 

I have the very same experience with 4.10 like Alex. HQPE won't see my Ultrarendu and I cannot get off the config screen either. It is working all fine with 4.9 tho.

[Miska]

I've done some fixes about NAA discovery, and bunch of other things. So I will make some new releases soon. I'll try to include some new code that is less allergic to order of powering things up when NAA is involved...

 

[koldby]

1 hour ago, Miska said:

 

Why do you have two? Anyway, you can collapse both into one...

 

I'll add more flexibility to the matrix config once I get over some of the new GUI hoopla...

 

I am sorry, but I do think that you do not under stand , how I have done it.

First I measure my speakers with REW and with that result I make a PRE-EQ. I put the pre-eq into Jriver and use the WDM driver in Jriver when I let audiolense measure the system. 

That means , Audiolense measures and make the convolution file with the pre-eq in the chain. 

So I need to have the PRE_EQ and the convolution in the playback software to get the correct result.

I am sorry if I fail to expålain this the right way, but I am NOT using two convolvers .

I use one pre-eq setting and then the convolver file from Audiolense.

So I need something in the HQplayer chain to accept the PRE-EQ settings before the convolver does its job , or else I do not get the right result.

 

Again I use the PRE-EQ before I measure with Audiolense , because this makes the room correction/EQ so much better than just using Audiolense to correct everything . You can read about this technique here: https://www.diyaudio.com/forums/full-range/242171-towers-25-driver-range-line-array.html

[DancingSea]

54 minutes ago, Miska said:

I've done some fixes about NAA discovery, and bunch of other things. So I will make some new releases soon. I'll try to include some new code that is less allergic to order of powering things up when NAA is involved...

 

 

Congratulations to Finland for winning the gold medal at the hockey world”s, and defeating Sweden, Russia and Canada along the way.  Surely a national celebration of this epic magnitude will inspire a HQPlayer 4 flash sale 🤣

[Mark80219]

Hi , I ues the latest version 4.03 and it seems a little unstable now , I open the main process then the client process , I found that it just randomly pick a cpu core and use it all.

And it goes down to 0 after I click and enter the settings page.

If I reverse the starting order of two process , then everything is fine.

Is this just me or it happens to you guys,too?

[Yviena]

@Miska If i want to use hqplayer without any upsampling using PCM, and using the digital volume control do i also set dithering to none or should it be set to TPDF if volume control is used?

[Miska]

1 hour ago, Yviena said:

@Miska If i want to use hqplayer without any upsampling using PCM, and using the digital volume control do i also set dithering to none or should it be set to TPDF if volume control is used?

 

Yes, you need to use some dither or noise shaper, otherwise you get quantization distortion. Also pay attention to having correct "DAC Bits" in case the DAC chip is not accepting 32-bit inputs, to avoid dithered LSBs from being cut-off and as result quantization distortion appearing.

 

TPDF is a good standard flat noise floor dither.

 

[Miska]

8 hours ago, koldby said:

I use one pre-eq setting and then the convolver file from Audiolense.

So I need something in the HQplayer chain to accept the PRE-EQ settings before the convolver does its job , or else I do not get the right result.

 

I'm not sure I understand why you do it that way. But in any case you can collapse both EQ's into one convolution filter.

 

[LoryWiv]

Jussi, my understanding is that HQ Player 4.03 Desktop has no specific MQA capabilities, so I assume the inclusion of poly-sinc-mqa filters is for those who output to an MQA capable DAC. My DAC is not, and I'm not very interested in MQA frankly, but I do have a few FLAC downloads with MQA encoding. 2 related playback questions:

 

1. Does HQ Player essentially ignore the MQA encoding and play the file at it's intrinsic bit rate / sampling rate (EG - 16/44.1)?

2. if so, is there any advantage to selecting the poly-sinc-MQA filters when playing these files output to a non-MQA DAC?

 

Thank you!

[Yviena]

56 minutes ago, Miska said:

 

Yes, you need to use some dither or noise shaper, otherwise you get quantization distortion. Also pay attention to having correct "DAC Bits" in case the DAC chip is not accepting 32-bit inputs, to avoid dithered LSBs from being cut-off and as result quantization distortion appearing.

 

TPDF is a good standard flat noise floor dither.

 

Hmm i see if i do have linearity measurement should i still use that to set dac bits or set it to what the xmos chip is exposing eg 24 bit.

[3ggerhappy]

First post!

 

Ok so the ryzen 3rd Gen was announced 5 hours ago with its specs:

 

https://www.anandtech.com/show/14407/amd-ryzen-3000-announced-five-cpus-12-cores-for-499-up-to-46-ghz-pcie-40-coming-77

 

will be released July 7

 

now question is which cpu could hqp fully utilized? i have hold on buying an i9 9900k waiting for this, also ryzen 3800x 8Cores 65W TDP or 3900x 12Cores 105TDP. Thoughts guys?

 

[koldby]

1 hour ago, Miska said:

 

I'm not sure I understand why you do it that way. But in any case you can collapse both EQ's into one convolution filter.

 

I thought I had explained why I do it that way: If I let Audiolense (or DRC) make the correction without making a soft and gentle EQ of the arrays, the result will be lifeless and artificial. The technique with first eq-ing the arrays and then let the Room correction software do the rest,  makes the end result so much better.

So how do I collapse the parametric settings and the wav. file from audiolense into one wav. file?

And still my question remains: Is it possible to let Jriver sit in front of HQplayer in the playback chain. That would solve the problem easily. 

Thanks for spending time on answering ..

 

[Miska]

5 hours ago, koldby said:

So how do I collapse the parametric settings and the wav. file from audiolense into one wav. file?

 

You convolve the impulse response with another one. IOW, you would filter the Audiolense's filter with the parametric EQ. When I find time, I'll add this kind of feature to HQPlayer so it'll happen automatically.

 

5 hours ago, koldby said:

And still my question remains: Is it possible to let Jriver sit in front of HQplayer in the playback chain. That would solve the problem easily.

 

It may work with UPnP on HQPlayer Embedded. Other alternative is to either use audio loopback to HQPlayer 4.

 

Or split the work to two computers and use for example AES between - this is what I'm doing between Mac Mini and HQPlayer Embedded computer.

[Miska]

6 hours ago, 3ggerhappy said:

now question is which cpu could hqp fully utilized? i have hold on buying an i9 9900k waiting for this, also ryzen 3800x 8Cores 65W TDP or 3900x 12Cores 105TDP. Thoughts guys?

 

There is almost no clock difference between 3900x and 3800x. I'm pretty sure I will try to get 3900x as soon as possible, it looks promising part with it's notably large caches compared to smaller models. Large caches can help HQPlayer.

 

And 3700x seems to be very potential for passive cooled configurations!

 

[micheloupatrick]

Hi Miska,

I was ready to upgrade to HQP desktop 4 but I have a short crackling/popping sound at the beginning of the playback : it 's actually quite loud (and seemingly independent from the volume setting) and sometimes happens when I stop the music too. Playback is otherwise fine but I can't upgrade for now with such an issue.

Everything was fine with HQP desktop 3 : source is Roon, DAC is a Luxman DA-06. Ubuntu 18.04 on an Intel NUC.

Thanks for your help,

Alex

 

[luisma]

1 hour ago, Miska said:

 

There is almost no clock difference between 3900x and 3800x. I'm pretty sure I will try to get 3900x as soon as possible, it looks promising part with it's notably large caches compared to smaller models. Large caches can help HQPlayer.

 

And 3700x seems to be very potential for passive cooled configurations!

 

that 3700x on a Streacom DB4 might work very well

[koldby]

2 hours ago, Miska said:

 

You convolve the impulse response with another one. IOW, you would filter the Audiolense's filter with the parametric EQ. When I find time, I'll add this kind of feature to HQPlayer so it'll happen automatically.

 

 

It may work with UPnP on HQPlayer Embedded. Other alternative is to either use audio loopback to HQPlayer 4.

 

Or split the work to two computers and use for example AES between - this is what I'm doing between Mac Mini and HQPlayer Embedded computer.

Thanks a lot Miska

[Miska]

3 hours ago, luisma said:

that 3700x on a Streacom DB4 might work very well

 

I'm thinking more about the equipment rack friendly FC10 which is rated for 65W TDP. Now I'm using it with Intel T-series 35W TDP CPU's and it gets barely warm when doing DSD512.