HQ Player

[ankur625]

hi, tried hqplayer few days back with my laptop as source, as currently i am not having my dac with me. after so many trials i liked sinc filter most. i upsampled cd quality to 192khz and it did wonders . i plan to use the sinc filter with my mojo once i get back the dac. anybody tried sinc filter to upsample ? most people prefer poly sinc varieties but to me sinc has still better clarity. can anybody tell in detail about this filter ? this sinc filter uses more resources than poly sinc but for a modern pc this is manageable .

[ankur625]

as a new user of hqplayer can anybody help me to get best sq upsampling setting for dsd64 to dsd256 for chord mojo ? i plan to use windows laptop with i7 4th gen . for i plan to keep pcm to pcm and dsd to dsd ( or dsd to pcm )as chord mojo does not handle dsd natively

[rickca]

No, the track is processed on the fly while playing back. Otherwise the waiting time for each track would be too long. For example if your CPU load while playing back is 50%, then pre-processing a track would take half of track's length. So a five minute track would take two and half minutes to process.

Thanks, Miska. This makes perfect sense now.

[tboooe]

No, the track is processed on the fly while playing back. Otherwise the waiting time for each track would be too long. For example if your CPU load while playing back is 50%, then pre-processing a track would take half of track's length. So a five minute track would take two and half minutes to process.

Curious, is this still the case if I use Roon and HQP? The reason why I ask is when I look at the Task Manager, I see a brief spike in network traffic when I start to play a track but then the network traffic settles to zero. I assume this is the track being downloaded from my NAS to my music server PC. It seems to me that perhaps Roon is downloading the entire track before sending it to HQP which then processes the file on the fly?

[Miska]

Curious, is this still the case if I use Roon and HQP? The reason why I ask is when I look at the Task Manager, I see a brief spike in network traffic when I start to play a track but then the network traffic settles to zero. I assume this is the track being downloaded from my NAS to my music server PC. It seems to me that perhaps Roon is downloading the entire track before sending it to HQP which then processes the file on the fly?

 

Yes, it is no different when used with Roon. I'm not sure what Roon does inside. It may also be the OS itself caching the file.

 

Also with plain HQPlayer playback, depending size of the file and amount of RAM in your computer, OS may decide to preload the entire file to disk cache (RAM) almost immediately. On my Macs I've had cases where this happened and OS X spun down the HDD which in turn caused lengthy gap when switching to next track because it had to spin up the HDD again (takes couple of seconds)... Most of the time this doesn't happen though. And it is possible to disable the HDD powersave feature in OS X preferences.

 

Modern operating systems use practically all the otherwise unused RAM as a disk cache.

[R1200CL]

Miska

 

Can't read the whole tread. Now that Roon is available on Qnap and Synology. Any plans of make HQPlayer available for these NAS. I would expect some versions of the NAS now has enough power.

[ted_b]

Miska

 

Can't read the whole tread. Now that Roon is available on Qnap and Synology. Any plans of make HQPlayer available for these NAS. I would expect some versions of the NAS now has enough power.

 

I would not follow suite. Keep Roon processing and HQPlayer processing away from each other. It is nice that Roon now officially supports the QNAP and Synology NAS machines (many have moved to them even prior to the announcement) but they would be the last place I'd want HQPlayer, especially given what HQPlayer requires (to do DSD512 upsampling, for example). Roon on your NAS, HQPlayer on your workhorse, and the NAA on your minimalist quiet isolated DAC-connected device. A nice audio menage a trois.

[craighartley]

. Roon on your NAS, HQPlayer on your workhorse, and the NAA on your minimalist quiet isolated DAC-connected device. A nice audio menage a trois.

 

Yup, that's what I got working today

[drabbis]

Miska,

 

I read on a article, your house is like mine, that your music room doesn't have ethernet available and the router is a on the floor above. Do you think streaming dsd512 via router to AP/bridge is possible? most wifi bridges are capable of 1200-1900mbps.

[rickca]

Most wifi bridges are capable of 1200-1900mbps.

You're not going to get anywhere close to that rate unless your wireless media bridge is very close to your router ... in which case you wouldn't need a wireless media bridge. Still a good question ... what rate is required for DSD512 when HQPlayer is sending data to NAA?

[bibo01]

You're not going to get anywhere close to that rate unless your wireless media bridge is very close to your router ... in which case you wouldn't need a wireless media bridge. Still a good question ... what rate is required for DSD512 when HQPlayer is sending data to NAA?

 

Around 22mbits/s

[Miska]

Around 22mbits/s

 

That's one channel...

 

So stereo DSD512 with 48-base rate: 48 x 512 x 2 = 49.152 Mbps plus of course few percents protocol overhead.

[Miska]

Can't read the whole tread. Now that Roon is available on Qnap and Synology. Any plans of make HQPlayer available for these NAS. I would expect some versions of the NAS now has enough power.

 

So far the best I've seen in a NAS is Intel Atom CPU. That's good for PCM upsampling, but for DSD it is not so good.

 

The only point in buying a NAS device I can see is ease of setting it up, otherwise those are scaled down computers running Linux server. So I rather buy a real server hardware and turn it into a NAS by installing Ubuntu Server or Debian Stretch. For Roon interoperability you need HQPlayer Desktop which in turn has a GUI so the computer needs to be graphics capable*. HQPlayer Embedded is a different thing that doesn't have GUI and thus doesn't need graphics capabilities either, but it is not Roon compatible.

 

Creating something like headless HQPlayer Server version would require reimplementing the application. Doable of course, but a lot of work...

 

 

*) When you run HQPlayer Desktop on Linux, it of course doesn't mean that the physical graphics output would need to happen on the same computer, thanks to X11 it can be redirected to some other computer. Like another Linux machine or Mac (there's a free Xserver for Mac). And it is also possible to Windows, but a bit more pain to set up or alternatively needs to go with a commercial solution.

[Le Concombre Masqué]

What kind of feedback is expected from beta 4?

cuda offload still is not implemented

convolution needs to be launched on fly and still does not remain checked

[Le Concombre Masqué]

What's so "hot" (pun intended) with 512 ?

 

I listened to my sacd rip of Bach's 81 GG Goldberg and fan noise was very noticeable about 1/4H before the end. With CUDA offload convolution ON and 64 to 128 cpu load was # 18% but it was so hot that there was a glitch. I relaunched with CUDA offload unchecked but found the sound harsh in comparison. CPU load raised to #38%...

 

I really do wonder if 512 provides pure benefits while it takes a toll on cpu... do wonder if 128/#15%cpu load (i.e.) is really beaten by 512/30+cpu load...

 

In my case I opted for CUDA offload + 128 + convolution as offering the best SQ but it's only with a CPU load of less than 18%. But I'm not that comfortable with 79°C at GPU sensor. Is this OK?

 

 

That's one channel...

 

So stereo DSD512 with 48-base rate: 48 x 512 x 2 = 49.152 Mbps plus of course few percents protocol overhead.

[tsaett]

Hello Jussi,

 

I would like to point you again to my suggestion to add a little selection in HQPlayer that enables one to choose a specific frequency given a specific source format.

I have just changed my 3,1GHz cpu for a 3,9 GHz cpu and am now able to upsample all PCM rates to 512 with Matrix Processing enabled and non -2s filters.

However, for DSD content the cpu is still too weak for 512, 256 however works.

All my problems would be solved if I could just check a box that says "512 for PCM" and "256 for DSD" or similar.

 

I would really appreciate a feature like this and I'm sure many others would as well, because it would enable them to use the best settings for all source material in their given hardware context.

I remember Geoffrey Armstrong suggested this to roon but I don't know how this is going and HQPlayer seems (to me) to be the more appropriate location for this.

 

Thanks!

Christoph

[drjimwillie]

I created two filters in Acourate and placed them in the convolution engine. How do I use them, please?

Do they start working as soon as I hit "OK"?

Do I have to check the "enable" box in the top left corner?

Do I have to do anything else?

Thank you

Bill

 

I love music & can get excited when I hear a song I like on my transistor radio at work, this does not stop me from striving to reproduce the live event within the limits of my system and room. I used to design radio studios and I know how to put a room together. While researching DSD DACs I learned about HQPlayer, I drank the Kool-Aid, I bought HQPlayer before I installed it. Mostly because someone helped me set it up. I had DSD128 coming out of my MacBook pro and I did not touch one setting, for fear that I would mess something up. Recently Small Green Computer built me a CAPS Pipeline with the new HDPlex ATX linear power supplies. Clean power through the entire computer. It sounds good, I would like to describe it to you but I have not plugged in the MacBook pro again or heard a standard Pipeline. I have been here from the beginning, keeping up with every post on this and the NAA thread. Recently someone posted that this is not easy, I accept that this is how things are when we are on the bleeding edge of a technology. The rewards are great! My advice, buy HQPlayer, work out the details later. We are blessed to have so many smart and passionate music lovers here willing to help us, thank you.

 

I created two filters in Acourate, another fantastic tool with a brilliant developer, who like Miska is there to help, unbelievable, I can not show my gratitude enough. When listening to the music, I was marveling over how great it sounded. When I went back into the convolution engine I noticed that the "enable" check box was not checked. I checked the box and got different, good sound. I searched to find out how to use the convolution engine and could not find much and what I did find was not rudimentary. Please see my simple question above.

Does the convolution engine start working as soon a I put in the filters and hit "OK"?

Do I have to check the "enable" box?

Acourate now allows you to create filters sampled to 352 kHz. I up sampling redbook and from roon/tidal to DSD128. Is this what I should I use?

Can I delete files I placed in the engine, I can replace them, should I delete them to see if it makes a difference?

What should I expect from the digital room correction, what is your experience, please?

Thank you,

Bill

[ted_b]

As per the instructions on page 25-26 of the HQPlayer manual, enable the IR filters, enable HF extension if your filters are lower sample rate than you playback, and set your gain appropriately. You can look in the log for actual safe gain settings (i.e if log says convolver is doing 6.3db then set gain for -6.3db to eliminate any limiting). Although convolver is on at startup (if enable checkbox is checked) you turn it on and off on main screen green checkmark icon (to do a/b listening).

[ted_b]

Hello Jussi,

 

I would like to point you again to my suggestion to add a little selection in HQPlayer that enables one to choose a specific frequency given a specific source format.

I have just changed my 3,1GHz cpu for a 3,9 GHz cpu and am now able to upsample all PCM rates to 512 with Matrix Processing enabled and non -2s filters.

However, for DSD content the cpu is still too weak for 512, 256 however works.

All my problems would be solved if I could just check a box that says "512 for PCM" and "256 for DSD" or similar.

 

I would really appreciate a feature like this and I'm sure many others would as well, because it would enable them to use the best settings for all source material in their given hardware context.

I remember Geoffrey Armstrong suggested this to roon but I don't know how this is going and HQPlayer seems (to me) to be the more appropriate location for this.

 

Thanks!

Christoph

 

Although checking DirectSDM will partially solve the "I want to upsample PCM to a different DSD bitrate than DSD sources" the idea of having this recommended flexibility is certainly a nice one. Geoffrey's idea of putting it in the comments tags is a workable one IMHO (workable from a user perspective, not discounting Jussi's heavy load of work ).

[drjimwillie]

Thank you Ted.

OK so if the enable box is not checked the filters will not do anything.

What I find confusing is that when I check enable, after I hit OK, I see a progress bar that looks like it is doing the convolution. And when I un-check the "enable" box and then hit the "OK", I see the progress bar again. Is it unloading the convolution?

 

I will look in the thread to remember how to start the log and where to find it. To match levels for comparison I remember that Acourate said that the filter is -8dB, so I have been setting the volume slider in roon for non-convolved files to -11 (-8 + the standard -3dB). Is that the same?

 

What was your experience/perception listening convolved files for the first time? After many tries Uli made me files to use in the engine. So at least I know that what I am using at the moment is good.

[tsaett]

Although checking DirectSDM will partially solve the "I want to upsample PCM to a different DSD bitrate than DSD sources" the idea of having this recommended flexibility is certainly a nice one. Geoffrey's idea of putting it in the comments tags is a workable one IMHO (workable from a user perspective, not discounting Jussi's heavy load of work ).

 

Hi Ted,

with DirectSDM the matrix processing would be disabled too, so a no go for me. For others this is certainly one option.

I'm not sure what you mean by putting it in the comments tags, is there already a solution existent?

 

Thanks!

 

 

 

@ Bill:

When convolution is active, the green check-symbol on top is active, when not it's not.

You can check and uncheck the convolution on the fly while listening by clicking the symbol, so you can instantly hear what the convolution is doing.

[tboooe]

I just came across a weird issue. I have HQP converting PCM to DSD128. However, I have any issues with any music files that have sample rates of 48, 96, or 192. Sample rates of 44.1, 88.2, 176.4 play fine. When I try to play these files with samples of multiples of 48 HQP will just flash a "Please Wait" dialogue box over and over. My CPU usage is fine (around 50%). If I uncheck Autorate Family option then everything plays fine.

 

BTW, my settings are polysinc mp -2s, ASDM7, 5644800

[Miska]

All my problems would be solved if I could just check a box that says "512 for PCM" and "256 for DSD" or similar.

 

This kind of logic gets to a point where most sensible solution would be to add scripting language hooks so the output format selection logic would become scriptable.

[Miska]

I just came across a weird issue. I have HQP converting PCM to DSD128. However, I have any issues with any music files that have sample rates of 48, 96, or 192. Sample rates of 44.1, 88.2, 176.4 play fine. When I try to play these files with samples of multiples of 48 HQP will just flash a "Please Wait" dialogue box over and over. My CPU usage is fine (around 50%). If I uncheck Autorate Family option then everything plays fine.

 

BTW, my settings are polysinc mp -2s, ASDM7, 5644800

 

Does your DAC support 48-base DSD rates in first place? What does the rate selection drop-list show in main window? The behavior would indicate that there is no suitable 48-base DSD rate available.

 

In addition, if you set the rate limit to 5.6 MHz, then the auto rate family cannot pick up 48x128=6.1 MHz but needs to stick with 3.1 MHz instead.

[Miska]

What kind of feedback is expected from beta 4?

 

If there are any new bugs (regressions) found...

 

cuda offload still is not implemented

 

Yes, because there is no production release for CUDA8. There likely won't be any CUDA-enabled beta, I'll release 3.14 when the CUDA8 production release becomes available.

 

convolution needs to be launched on fly and still does not remain checked

 

What does this mean in practice? I'm sorry but I don't understand what is the exact problem.