HQ Player

[Johnseye]

On 4/24/2017 at 6:07 PM, Dr Tone said:

 

poly-sinc-ext to PCM might be nicer though.  I have a thing for the ext filter but going to DSD with it won't work with the Ayre for 48kHz variant source.

 

Do you know why the PCM poly-sync-ext filter is limited to 44, 88 and 176?

[Johnseye]

9 minutes ago, Dr Tone said:

 

It's not, my DACs DSD support was.  Poly-sinc-ext is limited to power of upsampling.

 

Not sure I follow. HQ Player limits me to the same 3 frequencies for poly sync ext as well. 

[Johnseye]

I ran into a situation where I downsampled to PCM 48 and it locked something behind the interface into 48.  Even when I changed sample rate, the sample rate in the drop down menu it did not change the actual sample rate.  I had to uninstall and reinstall HQP to get things to work properly again.

[Johnseye]

On 11/4/2017 at 8:50 PM, Argopo said:

 

Spent some time this week listening, using ASDM7 only and various filters. With "auto rate family" checked and unchecked:

 

Slight static noise overlay with playback of 24 bit rate (48kHz family), 96kHz and 192kHz.

No static noise overlay with playback of 16/24 bit rate (44.1kHz family), 88.2kHz and 176.4kHz.

 

And by slight static noise I mean an infrequent barely audible pop that appears throughout a recording.

Playing back the same file with PCM up-sampling to 384kHz results in no noise.

 

My next test will be installing HQP on my workstation with AMD 1800x, 32GB, 1070 and 1080 TI for the CUDA cores and using that machine as my control pc to see if I can replicate the noise.

 

 

 

 

 

 

 

I've had the same issue when not using auto rate family upsampling 44.1 to 192.  Some filters have bad static and pops, some don't.  I'm guessing those filters can't handle integer multiple mismatches.  It's odd you've experienced that with integers that match.

[Johnseye]

4 hours ago, Miska said:

 

Those filter's don't allow you to do such ratios in first place. If you experience problems from 44.1k to 192k then there is something seriously wrong somewhere. Because that doesn't consume any notable amount of CPU time either.

 

What OS and backend are you using?

 

I was using 2016 Server. I'm about to rebuild with a new server. We'll see if I have the same issue then. 

[Johnseye]

On 11/10/2017 at 3:24 PM, Miska said:

 

Those filter's don't allow you to do such ratios in first place. If you experience problems from 44.1k to 192k then there is something seriously wrong somewhere. Because that doesn't consume any notable amount of CPU time either.

 

What OS and backend are you using?

 

Would this be buffer related, not processor?

 

I noticed another issue.  I have my dac bits set at 32 but HDPlayer is not upsampling to 32 bits.  My dac supports 32 bits.  Roon will upsample to 32 bits, if increasing the bit depth is even considered upsampling.

 

[Johnseye]

4 hours ago, Miska said:

 

Trying to figure out your signal chain... Are you sending to SMS-200 running as a NAA? In that case, leave HQPlayer buffer setting to "Default". How are you determining that HQPlayer is not upsampling to 32-bits?

 

 

I'm no longer using the sms-200. The chain is server with tX-USBexp USB card to a tX-USBultra to the DAC then preamp then amp....

 

The DAC uses the USB control panel to set ASIO buffer and streaming mode or latency. The server is running Roon and HQPlayer. 

 

HQPlayer is not displaying the upsample to 32 nor is Roon. 

 

[Johnseye]

3 hours ago, Miska said:

 

OK, then just leave HQPlayer's buffer time set to "Default" and it'll use what the ASIO driver proposes. In the ASIO driver set buffer size to maximum (maximum latency) and streaming mode to "safe" or "extra safe" or something like that.

 

 

OK, that is normal...

 

Is there any negative impact maxing the asio buffers?

How do I know HQPlayer is upsampling to 32 bits. I thought I had seen it displaying that before, but not now.

 

[Johnseye]

I'm probably doing something stupid, but just can't figure this out.  Why when I select SDM from the main screen do I get PCM selections?  My Holo Spring DAC is in NOS mode and I'm using Roon with HQPlayer.

 

[Johnseye]

Here's at 44.1x256

 

and 44.1x512.

 

[Johnseye]

1 minute ago, Bimmer100 said:

I’m late in joining on his, but are you using hqplayer with Roon?

 

Yes

[Johnseye]

3 minutes ago, Bimmer100 said:

It’s common mistake... but you must disable the Asio Holo audio driver within Roon as a source. Why? Because hqplayer and Roon cannot use the driver at the same time! Roon must use the hqplayer as the source. This should fix the problem 

 

Thanks Tim, that was it.  I would never have guessed.  Now I get to find out how upsampling to DSD512 is.

 

Walla!

 

[Johnseye]

2 hours ago, shadowlight said:

@Johnseye, the only time I have seem similar issue is when HQPlayer thinks the DAC only supports PCM (usually, when I forget to move the Amanero USB interface for the DAC from Linux based NAA and do not have DoP checked) .  Does HQPlayer logs show the DAC supporting DSD?  Can you also try HQPlayer outside of the Roon environment and see if you see anything different.

 

Tim explained it here, although I've never had a DAC compete with HQPlayer to where it needed to be disabled in Roon before.  A minor inconvenience, if at all.

 

 

[Johnseye]

On 8/16/2018 at 2:34 PM, Miska said:

 

If you want something really long that is like M-Scaler, then closed-form or poly-sinc-xtr would be closest.

 

 

Hey there Jussi, can you please explain the differences between poly-sinc-xtr, sinc-M and closed-form-M filters?

and by really long, do you mean taps?

[Johnseye]

1 hour ago, Miska said:

 

Yes, number of taps = length.

 

ext2 and sinc-M are similar, but with different length. xtr cuts a bit higher up (see plots earlier in the chain). All these go into apodizing category. While closed-form is strictly non-apodizing.

 

 

xtr has 1M taps in PCM but ext2 doesn't have that many correct?  Does sinc-M have 1M taps?  I've been assuming the M stands for million.

[Johnseye]

4 minutes ago, Miska said:

 

xtr and ext2 have roughly the same, for same conversion ratio, but they don't have any fixed number of taps. And yes, sinc-M has one million taps.

 

 

Ah ok, so no fixed number of taps for xtr and ext2.  Would they ever have over one million taps and is sinc-M a consistent one million taps?

[Johnseye]

I'm having an issue with HQPlayer Desktop 4 pausing music playback.  The backend is an NAA to Hugo TT2 DAC.

Music files stored locally or on the NAS exhibit the same behavior.  Filter, dither and buffer settings don't change the behavior. 

HQPlayer Desktop is running on a Windows 10 PC.  The NAA is running on Audiolinux.

 

Playback is pausing constantly but inconsistently.  Throughout a 3 minute song it could occur 20 times randomly.  It could make it 10 seconds into a song or 40 seconds before occurring the first time.

[Johnseye]

2 hours ago, Solstice380 said:

@Johnseye  Only times I’ve had that type of “pause” issue was when HQP couldn’t get the file fast enough due to either the drive went to sleep or network got messed up.

 

The drive the music is on is an Optane 905P.  Probably one of, if not the fastest drives available today.  To test I placed audio files on this drive which is also the OS.  It's not going to sleep.  The network interaction is something I'd like to know more about.  How does HQP4 interact with the NAA and what kind of network settings can cause issues?

 

2 hours ago, AnotherSpin said:

 

For some time I was comparing Audiolinux, Euphony and Miska's USB images for NAA endpoint in combination with HQP desktop on mac mini. There were some very rare pauses when using Audiolinux. I use Miska's NAA almost exclusively now because of several other reasons, but there are no pauses either. That would be my advice, to test Miska's NAA image instead of AL to check pauses behaviour.

 

That's my next plan, to install Miska's NAA on my endpoint.  I believe it's a Linux image as well.  I wonder if there's a Windows to Linux compatibility issue? I can also install Audiolinux or Euphony over the current Windows 10 install.  That would be the second step if Miska's NAA doesn't resolve it. Thanks.

 

[Johnseye]

53 minutes ago, AnotherSpin said:

 

Yes, it is Linux. Never heard about compatibility issue, and it shouldn't be if Miska specifically designed it to work with any OS server HQPlayer. Please be aware Miska's NAA images come in two ways. First is single NAA image and second is a part of HQP Embedded image. There might be some difference in SQ between two images. 

 

I am licensed for HQP Embedded, but using that on a different Audiolinux server.  It feeds the NAA in question without too much of an issue.  Every once in a while, usually never but maybe once a song, playback will pause but I always thought it was Roon doing that until now.  I never tested HQPlayer Embedded direct to the NAA with this server where Roon wasn't involved.

[Johnseye]

23 hours ago, AnotherSpin said:

 

Yes, it is Linux. Never heard about compatibility issue, and it shouldn't be if Miska specifically designed it to work with any OS server HQPlayer. Please be aware Miska's NAA images come in two ways. First is single NAA image and second is a part of HQP Embedded image. There might be some difference in SQ between two images. 

 

21 hours ago, bobflood said:

Having Roon and HQP on the same machine can indeed be problematic. There is another guy posting above with the same problem on a ST i5. I have both running on a Win 10 machine and it took some tweaking to get it to work without this issue.

 

8 hours ago, Solstice380 said:

Cross that one off the list!!!  LOL  I'm using NAA on Win10 with the Amanero ASIO without issue.  I don't have Roon on this machine, though.

 

I appreciate your help with my troubleshooting.  Last night I ran Jussi's NAA and had the same issue.  Because I'm using some custom switches with the etheRegen, SOtM sNH-10G and Buffalo BS-GS2016 I decided to see if one of these were the culprit.  I swapped them all out for standard Netgear switches I had on hand.  No pauses with HQPlayer.  I then added the other switches back in one by one.  It turns out the offender is the SOtM sNH-10G.  It didn't matter whether fiber or ethernet was used.  This switch caused consistent issues with HQPlayer.

 

Question to the community, is anyone else using a SOtM sNH-10G with HQPlayer?  @Miska have you come across compatibility issues with this switch and your software in the past?  This is not a cheap switch and it will be disappointing to have to stop using it because of this incompatibility. 

[Johnseye]

30 minutes ago, bobflood said:

Not with that particular switch but I did have a similar problem with a 10G port on a switch so I changed over to a 1G port on the same switch and no problems anymore.

 

I wish I could say the sNH-10G is a 10G switch.  The model number is misleading as it's only a 10/100/1000.

[Johnseye]

1 hour ago, Miska said:

 

Only switches I have are Cisco and Hewlett-Packard Enterprise (HPE), and one ZyXEL. No problems with these. With managed/smart switches it is important to check 802.3x is enabled for all ports.

 

It is important to check that all network infra supports 802.3x, 802.1p and optionally 802.3az standards.

 

My PoE powered WiFi access points are also from HPE/Aruba. NAA is working fine over these too up to 8 channels of DSD256.

 

 

Why are 802.3x, 802.1p and 802.3az important to HQPlayer?

 

[Johnseye]

On 2/12/2021 at 5:03 PM, Miska said:

 

I don't know about that (or very interested either), but they do accept DSD1024. Which is the same from USB point of view.

 

 

Jussi, is there a fix yet for PCM 1.536 and DSD 1024?  I just received a Holo May and I'm only getting PCM 768 @ 24 bit.  No 32 bit either.  While HQPe has options to select those settings the NAA isn't supporting it.

 

2021/04/17 20:03:04 NAA output connect to 192.168.1.6:43210 [ipv4]
2021/04/17 20:03:04 NAA output network endpoint: Holo Audio UAC2.0 Gen2 Enhanced: USB Audio (hw:CARD=Enhanced,DEV=0)
2021/04/17 20:03:04 NAA output discovered 1 Network Audio Adapters
+ 2021/04/17 20:03:04 NAA output connect to 192.168.1.6:43210 [ipv4]
2021/04/17 20:03:04 NAA output network format: 44100/24/2 [pcm]
2021/04/17 20:03:04 NAA output network format: 48000/24/2 [pcm]
2021/04/17 20:03:04 NAA output network format: 88200/24/2 [pcm]
2021/04/17 20:03:04 NAA output network format: 96000/24/2 [pcm]
2021/04/17 20:03:04 NAA output network format: 176400/24/2 [pcm]
2021/04/17 20:03:04 NAA output network format: 192000/24/2 [pcm]
2021/04/17 20:03:04 NAA output network format: 352800/24/2 [pcm]
2021/04/17 20:03:04 NAA output network format: 384000/24/2 [pcm]
2021/04/17 20:03:04 NAA output network format: 705600/24/2 [pcm]
2021/04/17 20:03:04 NAA output network format: 768000/24/2 [pcm]
2021/04/17 20:03:04 NAA output network format: 2822400/1/2 [dsd]
2021/04/17 20:03:04 NAA output network format: 3072000/1/2 [dsd]
2021/04/17 20:03:04 NAA output network format: 5644800/1/2 [dsd]
2021/04/17 20:03:04 NAA output network format: 6144000/1/2 [dsd]
2021/04/17 20:03:04 NAA output network format: 11289600/1/2 [dsd]
2021/04/17 20:03:04 NAA output network format: 12288000/1/2 [dsd]
2021/04/17 20:03:04 NAA output network format: 22579200/1/2 [dsd]
2021/04/17 20:03:04 NAA output network format: 24576000/1/2 [dsd]

As you can see, it is your NAA that only supports 24 bit and only up to 768000 PCM.

[Johnseye]

On 4/18/2021 at 12:57 PM, ted_b said:

I would think it is YOUR NAA, not Jussi's code.  The May needs the correct firmware (3.0.12) and the correct driver on your NAA.  Jussi's software does way above what you are looking for.

 

10 hours ago, Miska said:

 

Did you remember to flash USB firmware that has those rates enabled? I believe by default the DAC nowadays ships with the highest rates disabled.

 

 

It wasn't the NAA, it was the firmware.  Zeljko at Euphony thought it might have been the NAA.  After "downgrading" the May dac's firmware everything works as expected.  Thanks Jussi for pointing me in that direction.