HQ Player

[austinpop]

9 minutes ago, ericuco said:

For Scenario 3 (DSD to PCM), you need to change Default Output Mode from [source] to PCM. Otherwise PCM stays PCM and DSD stays DSD.

 

Thanks, excellent point. I was actually aware of that, but should have indicated correctly in the diagram!

[scintilla]

So, I did a thing.  To keep myself from spending some real $$$ on a Holo May, I bought a D90 to play with for DSD direct, since this is undiscovered country for me having had a PCM-only dac for the last 8 years.

 

If I am reading the datasheet correctly for the AK4499EQ, then DSD operates with just 44.1k base rates and not 48k rates too.  So, given that, am I better off to use [source] to leave 48k base rate PCM files in PCM to the DAC, or to have HQP output a 44.1x128 or x256 DSD to the D90 for 48k base rate files?  I am using a convolution filter at 192Khz, but I can always build a new one at a higher/different rate in acourate. Comments?

[giordy60]

On 2/8/2020 at 9:42 PM, Miska said:

 

You could download the bootable HQPlayer Embedded image (HQPlayer OS) from here: https://www.signalyst.com/custom.html

 

Boot it up (no installation needed, just a spare USB memory stick), configure and try playback with it. Alternatively you could use it as a NAA and see what kind of playback results you get that way. If playback works with it, then in current setup there is likely a software related problem. If it still doesn't work on the same hardware, then the problem is likely related to the hardware.

 

Since running HQPlayer Embedded this way is fully self-contained known environment, including the OS itself, it is good reference. This lifts many software related uncertainties.

 

Free trial mode is enough for this purpose.

 

Jussi

problem solved.....
the bottleneck was the 10/100 mediaconverters
in pmc 1.5 mhz there is a flow of over 100mbit of data per second.

thanks for the advice, I'll try embedded on a usb stick to understand how it works 👍

[Miska]

10 hours ago, austinpop said:

Scenario 1: PCM to PCM Upsampling

 

Only the following settings apply:

 

Yes, those and "DAC Bits" which defines word length output is dithered / noise-shaped to.

 

10 hours ago, austinpop said:

Scenario 2: DSD to DSD upsampling

 

Only these settings apply:

 

Yes.

 

In addition DirectSDM affects all scenarios. For DSD -> DSD doing raw bypass and for other cases locking volume control.

 

10 hours ago, austinpop said:

Scenario 3: DSD to PCM upsampling

 

 

Yes otherwise, except also "Filter Nx" and "6 dB gain" apply to this case.

 

[Miska]

11 hours ago, clang said:

Does the library allow duplicates? I have different pieces of music in an album(e.g. 2 concertos or a symphony and an overture) put into separate folders. When I search using Client, I can only get one but not the other. I didn't seem to have this problem in earlier versions of Desktop.

 

Yes it does, but depending on views, identical artist + album may make it not appear. I need to check this. This depends on the client side and not on the HQPlayer library itself. So for example HQPDcontrol's path view should likely still show both.

 

The way I distinguish between such cases myself is that I add some additional description to end of the album name in HQPlayer's Library editor. For example "(HiRes)", "(remix)", "(SACD)" or similar.

[Miska]

On 8/6/2020 at 6:09 AM, jamesg11 said:

Any chance that this new top of range iMac will cope with EC modulators & 48x256, plus 2 channel convolution?

 

”Configurable to 3.6GHz 10‑core 10th‑generation Intel Core i9 with Turbo Boost up to 5.0GHz”

 

Likely it does, that is probably i9-10900K. I have one with i9-9900K and it works. But with the caveat that the cooling fan can get pretty loud under higher loads! So likely not something you'd want to have running in the listening space.

 

[Miska]

On 8/5/2020 at 12:34 PM, Zauurx said:

Shark teeth

 

For my understanding: when I use HQP (PCM 705/768 >> RME ADI-2) the processor works in cycles.
6/8% for 2 sec then a rise to 23/25% and we start again.
We see the drawing in "shark teeth".
Is it linked to a buffer management of HQP or ROON / HQP?
I use Roon as a remote control.
Important detail, I also have an HAF convolution (via Matrix menu)

 

 

It is normal. Note that with those graphs you need to take into account that every data point in the graph is average load during the graph sampling (update) period. Then rest depends on how things happen in/out of sync between that resource monitor and HQPlayer. So it may look different at different times.

 

[firedog]

On 8/6/2020 at 6:09 AM, jamesg11 said:

Any chance that this new top of range iMac will cope with EC modulators & 48x256, plus 2 channel convolution?

 

”Configurable to 3.6GHz 10‑core 10th‑generation Intel Core i9 with Turbo Boost up to 5.0GHz”

I just got an i7 10700K PC (slightly slower) and it will handle it no problem. Doesn't even work hard. Even with PEQ also added in. Convolution is child's play for decent processors. Barely moves the CPU "needle".

The upsampling to DSD, and especially 2X and 4X is what stresses them.

I also bought large quiet fans and a "soundproof" case, and you can't hear it unless the room is quiet and you put your ears near it. 

[firedog]

@Miska

On another thread it was recommended to set CPU cores to run at a uniform clock speed as this improves SQ. Anything to that? And regardless of SQ when is that a good idea when using HQP.

[Miska]

37 minutes ago, firedog said:

On another thread it was recommended to set CPU cores to run at a uniform clock speed as this improves SQ. Anything to that? And regardless of SQ when is that a good idea when using HQP.

 

CPU clocks are not related to sound quality, apart from likelihood of drop-outs/stutter due to running out of CPU power.

 

Running all cores at same clock reduces clock boost capabilities of the high loaded cores and burns unnecessary electricity on the lower loaded cores leading to unfavorable power consumption and thermal load.

 

[Zauurx]

1 hour ago, Miska said:

 

Likely it does, that is probably i9-10900K. I have one with i9-9900K and it works. But with the caveat that the cooling fan can get pretty loud under higher loads! So likely not something you'd want to have running in the listening space.

 

Small is beautiful but for some filters, we need power ! 
Apart from the noise of the cooling fans .. what do you think of the "digital noise" and "electric pollution"  used by these fans?
Snake oil ??
Fanless is only for the last streamer ?

[Miska]

7 minutes ago, Zauurx said:

Small is beautiful but for some filters, we need power ! 
Apart from the noise of the cooling fans .. what do you think of the "digital noise" and "electric pollution"  used by these fans?
Snake oil ??
Fanless is only for the last streamer ?

 

Fans are not so big in terms on electric pollution, especially because they are powered by their own PWM controllers. For example on iMac bigger sources of noise are HDD, CPU, GPU, display itself, etc. I have not tested how clean the iMac USB is in this respect. If you use ethernet to connect a DAC through NAA or RAVENNA, you don't have to worry about this in first place.

 

Fanless is just about the acoustic noise. I have two fanless servers and three quiet servers with fans, all those are fine. If it cannot be quiet, then placing the server in a different room and using NAA is the solution. NAA should be always fanless (and of course low electric noise) because it doesn't need much processing power and thus cooling.

 

[kelvinwsy]

I am now 1st 2 cores on 4.1 ghz and the other 6 of my 10700k on 4 ghz .. will try to lower even more next . Vcore  down to 1.24 v , polysinc xtr mp/lp smooth !

I use a Cooler Master 360 AIO with 6 x12 cm fans in push pull config I have 11 fans in my Lian Li O11 glass PC case.. All fans are with Elfidelity fsn filters ..

3 x12cm Intake 2x14 cm exhaust fans are on a Separately powered NZXT Sentry 3 fan controller.. So are the 3 push.fans on the  AIO radiator!  This fan controller can do 15watts x 5 channels each 

Only the AIO PUMP and 3x 12cm pull fans on tbe AIO radiator are rinning off the Motherboard  connections ! The 3 pull fans hv half speed Noctua reisistors z Almost silent! The other 8 fans are soeed controlled from the NZXT Controller.. 

Lots of airflow .. At so reduced fan soeeds Almost Sikent operation..

 

SQ us SO Good ! my Cpu Temp is 52 deg and my VRM is at 36-38 deg celsius

[kelvinwsy]

Of course DSD 256 EC7 No GPU

[Miska]

I'm running all fans on the Fractal Design Define R case from the Gigabyte motherboard fan connectors. And all fans are set to Silent profile from BIOS, so they run at minimum speed necessary based on the motherboard temps.

 

[kelvinwsy]

The beauty of using a externsl dsn controller is that PWM fan speed adjustment pulses are minimized in their effects on the MB. Then i half speed the MB connected system fans ! That helps too. Finally the Elfidelity fsn filters also filters out power fluctuations. 

A fanless cade is my next build - waiting for Hdplex's yet to launch 500watt full Linear PSU 

[clang]

"Yes it does, but depending on views, identical artist + album may make it not appear. I need to check this. This depends on the client side and not on the HQPlayer library itself."

 

Grateful if you would do that and offer a solution.

[clang]

9 hours ago, Miska said:

NAA should be always fanless (and of course low electric noise) because it doesn't need much processing power and thus cooling.

 

So it is safe to disconnect the fan from my RaspberryPi4  acting as NAA?

5 hours ago, Miska said:

And all fans are set to Silent profile from BIOS, so they run at minimum speed necessary based on the motherboard temps.

 

 

Could you incorporate this in the NAA image?

[Miska]

2 hours ago, clang said:

So it is safe to disconnect the fan from my RaspberryPi4  acting as NAA?

 

I don't know... Would need to ask engineers who design the hardware...

 

2 hours ago, clang said:

Could you incorporate this in the NAA image?

 

It is not related to any OS, it is motherboard feature, mostly I think on IPMI/PMIC/PMU controller.

 

Quote

Grateful if you would do that and offer a solution.

 

At the moment solution is to edit album name in the suggested way, so that there's no naming clash in combined artist + album.

 

[Miska]

7 hours ago, kelvinwsy said:

The beauty of using a externsl dsn controller is that PWM fan speed adjustment pulses are minimized in their effects on the MB.

 

What effects? The non-beauty part is that there's no connection between individual fans and motherboard temperature sensors, unlike when you let the motherboard fan controller firmware run the fans. I don't see a problem with PWM fan speed adjustments, because all the motherboard voltages are controlled in similar way (CPU, etc) using PWM switching regulators. Fans are really low power and non-problematic devices compared to the low voltage high current things.

 

[ASRMichael]

Looking for some help/clarification. @Mitch has made me my 1st set of convolution filters, ranging from 44.1 -384khz. 
 

few questions to help get my head round things. I’m using Euphony running HQPlayer. 

 

- I’ve uploaded left & right files individually, for example. Went to HQplayer Convolution menu, attached 44.1khZ for left & right & pressed apply. Then repeated the process for the rest of the files. Is this right? Presume Convolution menu is the correct menu? 

 

- is my understanding correct here. HQplayer receive source file, for example a 44.1khz file, then makes convolution filter adjustments, then upscales to my desired choice? Can i still use DSD or does it need to be PCM.

 

- There’s a box for Expand HF in convolution menu, should I tick this? 

 

- How do you know if convolution is working? I had look at logs doesn’t say anything about convolution. 

 

 

[scintilla]

Selecting expand HF scales the filter for the appropriate rate, so yes tick that.  I don't know if Jussi enabled a set of rate families in the latest versions, so I stick with one set of convolution filters at the highest rate that I typically will desire in my case 192khz, or the highest rate available if outputting DSD.  HQP takes care of the sample rate conversions, so there's no need to worry about that and you can output in either DSD or PCM as desired.  One way to tell if you have your filters implemented properly and operational is if the output level is significantly lower due to the convolution. 

[ASRMichael]

13 hours ago, scintilla said:

Selecting expand HF scales the filter for the appropriate rate, so yes tick that.  I don't know if Jussi enabled a set of rate families in the latest versions, so I stick with one set of convolution filters at the highest rate that I typically will desire in my case 192khz, or the highest rate available if outputting DSD.  HQP takes care of the sample rate conversions, so there's no need to worry about that and you can output in either DSD or PCM as desired.  One way to tell if you have your filters implemented properly and operational is if the output level is significantly lower due to the convolution. 

Thanks, one question if you can answer? for 2 channel stereo should I use Convolution menu or Matrix pipeline?

[Miska]

13 hours ago, ASRMichael said:

Looking for some help/clarification. @Mitch has made me my 1st set of convolution filters, ranging from 44.1 -384khz. 
 

few questions to help get my head round things. I’m using Euphony running HQPlayer. 

 

- I’ve uploaded left & right files individually, for example. Went to HQplayer Convolution menu, attached 44.1khZ for left & right & pressed apply. Then repeated the process for the rest of the files. Is this right? Presume Convolution menu is the correct menu? 

 

You only need ones at 352.8k or 384k, the other ones are unnecessary. What ever is uploaded last is the effective one.

 

13 hours ago, ASRMichael said:

- is my understanding correct here. HQplayer receive source file, for example a 44.1khz file, then makes convolution filter adjustments, then upscales to my desired choice? Can i still use DSD or does it need to be PCM.

 

Yes. It also works for DSD.

 

13 hours ago, ASRMichael said:

- There’s a box for Expand HF in convolution menu, should I tick this?

 

Only if your filters are lower sampling rate than let's say 176.4/192k. If they are higher than 44.1/48k they should have frequency response extending to Nyquist of the filter sampling rate - usually flat from about 20 kHz on. For example Acourate can create such when "Brickwall extension" is enabled. If not, you should use filters that have proper response up to Nyquist of the filter sampling rate and enable Expand HF option.

 

13 hours ago, ASRMichael said:

- How do you know if convolution is working? I had look at logs doesn’t say anything about convolution.

 

If you have "Enabled" checked on the /convolution page, then it should be enabled. Also in the log file it should say "Convolution enabled".

 

[ASRMichael]

@Miska sorry my lack of knowledge here is showing, re expand HF. I sampled using mic at 48khz. Then sent files to @mitchcohe created full set of files from 44.1-384khz. (Using Acourate). I’ve uploaded the 384khz to Convolution menu. 
 

Not sure if relevant but sample rate was from 10hz-24khz. 
 

based on info above should I tick the expand HF?

 

Cheers