HQ Player

[SwissBear]

Hi everyone,

 

New to this thread. Just made the step from Jriver to HQPlayer with the goal to use room correction convolution in native DSD. Very happy with what I have been hearing so far. Congrats to Miska

 

I have a question in relation to DACs. Mine is limited to DSD128, and probably down-sampling internally to 176.4 for processing. So I will contemplate changing. I have read that there are some users of Merging+Nadac DAC here. Is there a common view that this is the best DAC for making the best out of HQPlayer or are there giant killers which would be worth considering at a lower price ?

[SwissBear]

Disclaimer: I have bought a Merging NADAC but I have not yet heard it in my system. I have been told that it will arrive next week or the week after.

 

The reason why I bought a NADAC was because it is the best DAC that offers 8 channels of conversion, without going to silly money. IF I had the money, I would be using an MSB Theatre 8 DAC, but that costs USD$125,000.

 

If you only need 2 channels of conversion, and you want to take advantage of HQPlayer's ability to upsample to DSD512, and you don't need the networked audio feature of the NADAC - there is the T+A DAC 8. I believe this sells for about $3500 - cheaper than the NADAC.

 

Thanks Keith for the answer

[SwissBear]

Depends how many channels you need. For stereo, T+A DAC8 DSD (Windows only)...

 

Thanks Miska

[SwissBear]

Hi,

I have been using HQPlayer for a few months. I am currently running it together with Roon on a MacMini Late 2014 with i7 processor. My setup is MacMini -> SOtM sMS-200 -> Mutec MC3+ USB -> Devialet D900. HQPlayer is used to upsample eveything in 24/192. I am very happy so far.

I have read very enthusiastic reports of users upsampling to DSD 512. Having said that, Miska often pointed out the fact that upsampling from Redbook, the first step (I guess from 44.1kHz to 88.2 or 96kHz) was the most audible. So my first question to the community here would be: is there a real upside to buy a T+A DAC in order to be able to upsample from PCM 24/192 to DSD 512 ?

If the answer to this first question is positive, I understand that the T+A DAC does only support Windows as an NAA renderer. Miska has given the reference of the machine he is using in this field.

Remains the server with many interesting comments in this thread about the ideal dimension of the processor.

It has recently been writen that i7-6700K was requested ie TDM = 91W. Which brings me to the second question. Considering all what has been written on the topics of the importance of removing noise and using very low noise and low impedance LPSU in other threads, how important is this to the QS when building a server running HQPlayer ?

Do we absolutely need to go fanless, and have the best of breed LPSU on the server too, or is the quality of LPSU most important for the NAA renderer only ?

Finally, is there a difference in SQ when the server is running Windows vs Linux ?

Thanks in advance

[SwissBear]

18 minutes ago, bogi said:

The higher sample rate you can go with HQPlayer, the greater benefit you can gain. HQPlayer does upsampling in one tap and that's valid also for 44.1k (or any other PCM) to DSD512 (or any other DSD rate).

Thanks bogi.

I understand that HQPlayer is upsampling in one tap. My question was probably not well formulated. I meant: if upsampling from 16/44 to DSD 512 is a 100 segment, how big are the 16/44 to 24/192 resp. 24/192 to DSD512 sub-segments ?

Are we talking 50/50, which would mean to I am contemplating a further considerable improvement compared to what I have, or 80/20 or even 95/5 in which last case I should not consider this further investment.

[SwissBear]

16 minutes ago, ecwl said:

My take is no, not because of PCM 24/192 vs DSD 512 but because you're using Devialet D900. Even if you're to get the T+A DAC and play DSD512, the analog signal from the T+A would be sent to the Devialet and my understanding is that the analog signal would be converted back to 24/192 by Devialet's ADC and then reconverted back by its DAC for playback. I maybe mistaken on this point. But if I'm right, you're really getting no benefits from the DSD512. Definitely something worth checking out before diving into T+A. Obviously if you're going to replace the Devialet with a different amp to pair with the T+A, that's a different story and then it might be worthwhile. 

Thank you ecwl.

You have a good point here. This is the way the Devialet is handling the signal.

Having said that, some users of Devialet have invested in other DACs, totaldac for instance, and reported positive impressions. But these were PCM 24/192 DACs and not DSD ones. Not sure is this is very different as the DAC is producing an analog signal. Thanks anyway.

[SwissBear]

5 minutes ago, bogi said:

Hopefully makes sense for you.

Yes it does. Thank you very much for your time

[SwissBear]

54 minutes ago, ecwl said:

My understanding is that the analog signal would be converted back to 24/192 by Devialet's ADC and then reconverted back by its DAC for playback. I maybe mistaken on this point. But if I'm right, you're really getting no benefits from the DSD512.

Just a short complement to my previous answer: the Devialet is not a digital amplifier but a hybrid one. I will not develop further as this is not the topic here, but if you are interested, you can have a look at the technology here: https://vimeo.com/93141227

[SwissBear]

17 minutes ago, ecwl said:

Except the Devialet doesn't use analog volume control, it controls the volume digitally. This is the reason why all analog signals through the Devialet has to be converted via the ADC into 24/192 first and then apply volume attenuation digitally and then put through the internal DAC to feed the hybrid ADH amplifier. I concur that this is off-topic. 

Checked the configurator and it seems you are right. Thanks for the heads-up. The video is misleading  

[SwissBear]

Hi Miska,

Trying to plan for my next move:

• NAA will be on a small fanless computer, with a very low noise/low impedance LPSU
• server will be outside the room : how important is it to go fanless for the server ? Is it better to have a powerfull server with a good GPU card or to put more money in a fanless one with expensive LPSU ?

[SwissBear]

2 hours ago, Keith_W said:

 have a question for you - should I go with Broadwell-E (more CPU cores, but less pipeline speed), or Kaby Lake (cheaper, fewer cores, more pipeline speed). Essentially, does HQP prefer more cores, or more pipeline speed? 

 

Of course I have already factored a GTX 1080 video card into the equation. 

 

(Choice of Broadwell-E vs. Kaby Lake also means that I have more TDP to deal with, which in turn will impact on the choice of case and design of the PC, so this is a very important question for me!) 

 

Hi Keith,

This link might give you some clues: https://www.computeraudiophile.com/blogs/entry/493-building-a-new-server-for-multichannel-processing/

[SwissBear]

10 hours ago, Traktorist3d said:

Hello. I'm new here, please do not judge if such questions have already been. I have a few questions.
Dear Miska:
1.With what settings in rephase is it most optimal to build a filter for HQP?

Do I need to consider the FFT length and the number of tapes? As I understand it, tapes should be taken into account when we cut low frequencies.

 

 

2. Is it better to set temporary time delays through tools -chanel ballance? Or are there other ways?

3. I am using VB -audio cable for audio streaming. Is it possible to count on the maximum sound quality in this mode? Or transformations affecting the sound are inevitable?
I don’t understand why I have 16-bit input and output in VB - Audio, but for some reason the output is always 24 bits, and in HQP it is always 32-bit input. I understand that if dithering is applied at 16-24-32? Will this theoretically affect the sound quality? I don't seem to hear the difference, but for perfectionism I would like to know that the signal reaches HQP without changes.

It is clear that if the input to VB-audio 16 and output 24, this is not a question for you, but for the VB developer. Is it normal that HQP always has 32-bit input?

4. I would very much like to make it so that when the signal is fed to the input device, it automatically reproduces it. As in the case when with ROON we apply for HQP.

In general, the sound quality from HQP is very pleasing. The only thing for channel division of frequencies in the car, I would like to input from other players. For example when watching YOUTUBE, etc. With Virtual Cable, this works fine, but is sorely lacking in automatic playback when fed to an HQP input device.

 

 

 

What is nice with rePhase is that you see how many taps you need in the interface.

For instance, if you want to apply a reject low filter at 20 Hz and you do not have enough taps in your impulse, you will see the blue line which shows what you expect and the red line which shows what you get.

 

If you increase the number of taps, you can see that both curves overlap.

 

 

Hope this helps :)

[SwissBear]

For those of you who are interested in the performance of the Mac Mini M1 running HPLayer under Rosetta emulator: the limit I found is:

- DSD 64 to DSD 128 with ASDM5SEC or ASDM7EC, regardless of convolution

- PCM to DSD 256 with ASDM5SEC or ASDM7EC, regardless of convolution

Processor load is close to 60% including system, and memory pressure is reasonable

 

 

So there is hope to have a reasonable configuration for a stereo system when Miska has optimized the code for the new platform...

[SwissBear]

20 minutes ago, Mahler and Bach on Computer said:

@SwissBear, thank you for the update for this important new development.  Curious how the fan responded under the condition of DSD256.  Also, what happened you tried to do DSD64 to DSD 256?  

The fan kept very quiet anytime :-)

What happened when trying DSD 64 to DSD 256 was like a drop of network frames from time to time (blanks in the music). It might not be the CPU, but could be the network adapter which was overloaded, busy getting the music files from the NAS and sending the upsampled file to the NAA.

For now, the upsampling to DSD 128 seems to work for all music files, including PCM, DSD 64 and DSD 128. Even with ASDM5EC and ASDM7EC, which I could never make happen even with my very old but very powerful MacPro with Xeon processors...

So I'm more than happy for the time being, and very optimistic on the possibilities of this new machine...

[SwissBear]

3 hours ago, Mahler and Bach on Computer said:

I am thinking to buy Mac Mini M1 to replace my 7-year-old MacBook Pro.  Your information/experience is very helpful.  It is good to hear that fan keeps quiet all the time.  Given that there was no issue under the PCM-DSD256 scenario, it makes sense that you hypothesized that issue with DSD64 to DSD256 likely related with NAS file transfer.  Given I have been using NAS myself, I understand it well.  One way to get around this problem is to use some of those apps developed by @Geoffrey Armstrong, which generate files locally at the computer which running HQPlayer4Desktop.  This way HQP4D plays a locally freshly generated files, be it FLAC or DSD64, likely on SSD.  Personally I think it helps SQ as well without measurements, because power supply to NAS likely is not relevant any more for SQ.  

Thanks for the tip. Drawing conclusions from my tests, in the absence of any obvious bottleneck, seems difficult for me.

I will wait for Miska to release a native version of his software, compiled, and possibly optimized, for the new platform.

BTW, please do not take my tests at face value and make your own evaluation. In particular, it seems that upsampling to DSD 128 is pretty reliable and that upsampling to DSD 256 is pretty unreliable, whatever the original file format.

Just my 2C :-)

[SwissBear]

Hi everyone,

 

A short update on my experience with the M1.

 

First, I was intrigued by this post of Miska:

and I decided to measure the tension on the STP cable I had on my Mac Mini M1. And this was 30V :-(

So there seems to be a serious leak of current from the PSU on this device... which makes me expect the kit of @Superdad next year... This led me to generalize the use of UTP cables at home. Thanks @Miska

 

Otherwise, the listening experience with HQPlayer is very smooth. On most music files, the combos DSD256/ASDM5EC and DSD128/ASDM7EC are very smooth. I am still very much looking forward to the native version, but I am quite impressed by what I am hearing so far...

[SwissBear]

Whaoo Jussi ! Congratulations ! The Mac Mini M1 is able to upsample DSD 128 to DSD 256 with ASDM7EC and apply convolution.

And switching from one song to the other is so quick and smooth :-) :-)

I can't believe it !

Sincere gratitude for making this happen so quickly.

 

[SwissBear]

Hi everyone,

 

To answer a few questions:

- I mentioned upsampling D128 to D256 was flawless as this was the most demanding operation in the emulation mode

- upsampling from DSD 64 to DSD 256 with ASDM7EC is flawless too...

- my convolution is very demanding (524 ktaps impulses at 192 kHz) as I have to use a very steep filter to prevent modes in my room

- consumption of resources is shown here:

 

 

 

 

 

So congratulations and thanks again to Jussi :-)

[SwissBear]

Hi Jussi,

With all due respect and without any attempt to contest your position which is not to validate the Mac Mini M1 as a viable option for HQPlayer oversampling to DSD 256, I would like to share my observations:
- I have been listening to various input formats ranging from redbook to DSD 64 with the following setting:

- I noticed a kind of “white noise” a few seconds after changing input format. When restarting the track, the restitution was exempt of any glitch for the following tracks
- I would like to precise that my Mac Mini is completely headless (no screen no keyboard), with Screen Sharing not used, and Activity Monitor not launched. Only two applications running: Roon and HQPlayer
- I tend to consider that oversampling to DSD 256 is viable under the condition that input format is stable
- Oversampling to DSD 128 is viable whatever the changes in input format…
Hope this helps users to decide for themselves.
Pierre

 

[SwissBear]

1 hour ago, Miska said:

 

The difference here to my settings is that "Multicore DSP" is set to checked (force-enable all parallelizations), while I use it set to grayed (auto-detection). And I have poly-sinc-ext2 also for Nx rates.

 

Tomorrow I'll check also with "Multicore DSP" checked what kind of results I get.

 

Also Mac Mini and NAA have fixed IP address...

[SwissBear]

I spent the whole evening yesterday listening to music with HQPlayer in DSD 256. This was a great experience :-)
I have to admit that I have been facing some small glitches, mostly clicks and pops, during this time. These were random, very seldom, and there were also very few periods of white noise. But compared to what other processors have produced in this field in my environment, I still consider this M1 machine as promising.
I can only dream that we will find a way to tune the system and turn unnecessary services off (https://gist.github.com/pwnsdx/1217727ca57de2dd2a372afdd7a0fc21#gistcomment-3467443), or be able to boot in a near future from an ARM 64 Linux USB device, with HQPlayer on it (https://forums.macrumors.com/threads/ubuntu-linux-virtualized-on-m1-success.2270365/).
In the meantime, as Jussi has explained, DSD 128 remains the production environment, and DSD 256 more of a teaser for future developments.

 

[SwissBear]

11 hours ago, Uni said:

 

@SwissBear  Can you try to set SDM oversampling to poly-sinc-ext2 (not 2s) and check the performance and see if the fans kicks off?

 

I just tried. I kind of noticed that a 400 % load of HQPlayer was the threshold when clicks and pops started. And this is what happens with such oversampling filter. I did not stay long enough to evaluate fan behavior as this was not so pleasant to listen to  :-)

[SwissBear]

25 minutes ago, Miska said:

Some cases put the CPU load close to the maximum, but it still works and fan doesn't get too loud

 

Did you also get cracks and pops ? White noise ? Any idea how to get rid of that ?

[SwissBear]

14 hours ago, Miska said:

 

None, everything completely smooth. NAA is UpBoard running my NAA image and DAC is Holo Audio Spring.

 

I was able to replicate this behavior and isolate the source of the cracklings in my installation.

The fact that we have an "off-the-shelf" solution for less than 1'000 USD including NAA, to run DSD 256 in stereo with the best filters/modulators available, seems to me a matter of celebration.

Thanks for this native ARM 64 code :-)

[SwissBear]

33 minutes ago, The Computer Audiophile said:

Just fired up my wife’s M1 MacBook Pro to test HQP 4.8. DSD256 using ASDM7EC was a no-go. 

Did you check this box ?

 

See: