HQ Player

[Le Concombre Masqué]

thank you. I will stop searching for an issue on my side... I could live with Closed Form ; wonder if anyone has a clue to when using the poly sinc family is a must (because of aliasing ringing etc of the source).

 

happy new year all !

Hope this helps...

 

I'm using 3.12 with Yosemite on a Mac Pro 6,1 (quad core) and it works flawlessly with all flavors of poly-sinc*-2, poly-sinc-ext, and closed-form at DSD256 to an iDSD Micro.

 

I don't know if this is an issue, but I'm unable to use any of the non "-2" poly-sinc filters (with "Pipeline SDM" enabled or disabled). HQPlayer just sits there when I attempt to start playback (EDIT: I was impatient, playback starts eventually).

 

 

EDIT: I experimented a bit with "poly-sinc" (not the "-2s" version).

 

With "Pipeline SDM" both on and off

 

All tests with ASDM7

 

DSD256: Takes 30 seconds or more to start, solid stuttering

DSD128: Takes 20 seconds to start, sort of plays without stuttering but is unstable and the slightest additional load on the CPU will cause stuttering and clicks. The "top" command here shows CPU usage of HQPlayer is 215% (using over two cores constantly)

 

Using "poly-sinc-2s", DSD256 is solid on this Mac. The "top" command shows CPU usage of HQPlayer to be 145% (using less than 1.5 cores).

[Miska]

I would think this assessment spots the issue, except for one problem: for the past week I've only had HQP set to upsample all material to DSD 64 with either poly-sync-shrt-mp or poly-sync-shrt-mp-2s. My understanding is that there shouldn't be the type of issue you are describing under this scenario. Am I wrong?

 

Yes, those can convert any rate to any other rate, so conflicting settings cannot cause HQPlayer to stop.

 

Different aspect is that when there is a change in the source format, Roon will need to add a new stream to the playlist before the currently ongoing ends. This makes the transition to happen automatically. If the playlist runs empty, HQPlayer will stop and then Roon need to ask HQPlayer to start playing again once there is more content added to the playlist.

[Miska]

Filter 1: ON LS Fc 59.8 Hz Gain 10.0 dB

Filter 2: ON PK Fc 132 Hz Gain -8.0 dB Q 14.14

Filter 3: ON PK Fc 840 Hz Gain 2.0 dB Q 10.61

Filter 4: ON PK Fc 4,220 Hz Gain 16.0 dB Q 10.61

Filter 5: ON HS Fc 11,625 Hz Gain 12.0 dB

 

The notch filters are high-Q so not so strongly audible. The total gain of the filters is high, so you need to apply -16 dB gain compensation in HQPlayer to keep all peaks below 0 dB. So keep eye on the "Limited" counter in HQPlayer while playing to make sure it never increments from 0.

[bibo01]

Those already exist, those are volume up/down keys! And for tracks there are the corresponding track control keys.

 

For example Logitech K200/K270 keyboard has some of those keys (four, Play/Pause + Mute + Volume Down + Volume Up). K350 has also track control keys.

...

Sorry Miska, do we have problem to understand each other in 2016?!

I told you that the up/down keys work only after selecting the Volume knob!

 

I understand that with certain multimedia keyboards volume control is separate, but with all notebook keyboards is not.

 

My request stands.

[AnotherSpin]

There is a problem which happens from time to time. After playing an album and brooming it out from playlist the new album dragged into is not starting to play. The solution is simple - cleaning the playlist, closing HQP, starting it over again, dragging the same album back, now it plays. Not a big deal, and happens not every time, still something is not as smooth as with pre 3.12 builds. Yosemite, mac mini, closed form.

 

Update: the problem persists. Closing and opening HQP eliminates it, but there seem to be easies solution as well - clicking several times on a 'broom' icon seem to help as well.

[dean70]

Is the limited counter reported anywhere with the API? I use the replay gain tags in the track metadata to adjust the volume, but it is a bit of guesswork as to whether it is too much attenuation or not enough.

[Miska]

Sorry Miska, do we have problem to understand each other in 2016?!

I told you that the up/down keys work only after selecting the Volume knob!

 

Maybe we have something I'd call alignment issue... ;D

I didn't even know that up/down keys also work because I never tried and it is not part of my code at all. In my code there's bindings only for the multimedia keys. Probably this is one of the items that doesn't even necessarily work the same way on all platforms. I could add some other key bindings, but overriding the default OS/toolkit bindings is tricky and could lead to unexpected side-effects.

 

I understand that with certain multimedia keyboards volume control is separate, but with all notebook keyboards is not.

 

OK, I see, we need to then find some other keys that are widely available and are not already bound to some OS specific use. Mac OS X is good example of platform where annoyingly Apple has hard-wired bindings for multimedia keys, so even though I have bindings for those, the OS itself is stealing those and launching iTunes instead (unless you hack the OS not to).

 

How about Apple keyboard key bindings?

F7: Previous

F8: Play/Pause

F9: Next

F10: Mute

F11: Volume Down

F12: Volume Up

[madfloyd]

and then click on play in Roon again.

 

Sometimes just clicking on play in Roon again will restart the music.

 

 

I am never able to get music playing again via Roon iPad app. There is no 'chain' icon in the volume widget (perhaps there will be when the new app is released by Apple?) and I can hit Play a zillion times and music will never start.

 

So I have to remote desktop into my headless PC that runs HQPlayer and kill it and restart it.

 

It doesn't seem to happen only when a playlist ends or source format changes.

[Miska]

Is the limited counter reported anywhere with the API? I use the replay gain tags in the track metadata to adjust the volume, but it is a bit of guesswork as to whether it is too much attenuation or not enough.

 

Yes, it is part of the status() signal, last argument.

[dean70]

Yes, it is part of the status() signal, last argument.

 

ah the "clips" value - it was right in front of me Thanks.

[trcns]

""

But it still goes to the beginning of the play list. Who can I tell the program to continue from it left and not to start from the beginning?

 

Current state save doesn't store current track information, but I can add it on the next version...

""

 

I am wondering if this issue was addressed in this new release?

[1markr]

Stuttering problem (with all poly-sinc filters, @ DSD256) solved!

 

This is on my Macbook Pro (mid-2012, I7 - 2.3GHz, 16Gb ram). It was due to the Nvidia CUDA drivers. Even though my Macbook has a secondary GeForce GT 650M video card built-in, it was the CUDA driver(s) that were causing the stuttering with my poly-sinc filters using any modulator, and at DSD128 and DSD256 (and prob DSD64 as well, but I can't remember)

 

I built a new, clean bootable partition on a USB drive using El Capitan, my exaSound ASIO drivers, and HQPlayer, and the Nvidia CUDA drivers. Everything else loaded was just the default of the El Capitan install.

 

Just turning "off" the "CUDA" offload within HQP wasn't enough, and just removing the Nvidia prog from the System Preferences pane wasn't enough. I used EasyFind 4.9.2 (freeware) to search and remove all instances of Nvidia and CUDA.

 

Not sure if anything else caused my stuttering issue, but at least the clean install of El Capitan without CUDA drivers did the trick!

 

I am happy to enjoy the poly-sinc family of filters again at DSD256! Happy New Year!!

 

 

 

 

Btw, it's not just the ASDM7 modulator that doesn't work with the poly-sinc filters, but none of the modulators prevent are stutter free. All the poly-sinc filters (non -2s) stutter with all the modulators.

[Le Concombre Masqué]

hmm if I read you well you solved issues appeared with 3.12 by loosing the benefits it brings...

Stuttering problem (with all poly-sinc filters, @ DSD256) solved!

 

This is on my Macbook Pro (mid-2012, I7 - 2.3GHz, 16Gb ram). It was due to the Nvidia CUDA drivers. Even though my Macbook has a secondary GeForce GT 650M video card built-in, it was the CUDA driver(s) that were causing the stuttering with my poly-sinc filters using any modulator, and at DSD128 and DSD256 (and prob DSD64 as well, but I can't remember)

 

I built a new, clean bootable partition on a USB drive using El Capitan, my exaSound ASIO drivers, and HQPlayer, and the Nvidia CUDA drivers. Everything else loaded was just the default of the El Capitan install.

 

Just turning "off" the "CUDA" offload within HQP wasn't enough, and just removing the Nvidia prog from the System Preferences pane wasn't enough. I used EasyFind 4.9.2 (freeware) to search and remove all instances of Nvidia and CUDA.

 

Not sure if anything else caused my stuttering issue, but at least the clean install of El Capitan without CUDA drivers did the trick!

 

I am happy to enjoy the poly-sinc family of filters again at DSD256! Happy New Year!!

[bibo01]

Maybe we have something I'd call alignment issue... ;D

I didn't even know that up/down keys also work because I never tried and it is not part of my code at all. In my code there's bindings only for the multimedia keys. Probably this is one of the items that doesn't even necessarily work the same way on all platforms. I could add some other key bindings, but overriding the default OS/toolkit bindings is tricky and could lead to unexpected side-effects.

 

 

 

OK, I see, we need to then find some other keys that are widely available and are not already bound to some OS specific use. Mac OS X is good example of platform where annoyingly Apple has hard-wired bindings for multimedia keys, so even though I have bindings for those, the OS itself is stealing those and launching iTunes instead (unless you hack the OS not to).

 

How about Apple keyboard key bindings?

F7: Previous

F8: Play/Pause

F9: Next

F10: Mute

F11: Volume Down

F12: Volume Up

 

Apple keyboard key bindings are fine, as long as volume keys are for volume only and respond to volume only without previous selection.

Thanks

[firedog]

I am never able to get music playing again via Roon iPad app. There is no 'chain' icon in the volume widget (perhaps there will be when the new app is released by Apple?) and I can hit Play a zillion times and music will never start.

 

So I have to remote desktop into my headless PC that runs HQPlayer and kill it and restart it.

 

It doesn't seem to happen only when a playlist ends or source format changes.

 

Okay, I'm using the Windows Roon install as controller only. So apparently it is different.

There is supposed to be a new app version for iOS any day now, maybe that will solve some of your issues.

 

My understanding from the Roon community (forum) is that the whole Roon - HQP integration is not going to work as it should with an iPad until the new iOS app comes out. You should check at community.roonlabs.com.

[Miska]

ah the "clips" value - it was right in front of me Thanks.

 

Yeah, I regret naming of the attribute, because it is sort of wrong, but one cannot change the protocol once it has been defined...

[Le Concombre Masqué]

Am I right to claim that only Miska has a satisfying complete experience with 3.12 ? For the time being I simplified my setting to no NAA no Ram disk for apps while I switch between 3.11 & 3.12 depending on my choice of poly sinc or closed form filter...

[Solstice380]

Am I right to claim that only Miska has a satisfying complete experience with 3.12 ? For the time being I simplified my setting to no NAA no Ram disk for apps while I switch between 3.11 & 3.12 depending on my choice of poly sinc or closed form filter...

 

My Mac Mini is OK on Yosemite, mostly, using all modulators / filters to 128. Late 2012 i7 quad, 512 GB SSD, . Very lightly optimized OS. HQP runs 44.1 aiff files in Poly variants to 128 at around 1 core and overall CPU user load is ~15%.

 

I have one issue that I think is caused by the Mini and one issue with Closed Form. The Mini caused issue is a stutter and then drop out of the USB (direct connected to DAC) for a second or two then everything resumes OK. Only happens once per listening session and it can happen in the first minute or many tracks later.

 

The issue with closed form is going to 128 where at always about 2 minutes into a track the CPU use climbs to over 300% and then chokes! Going to 64 seems to work for a long time.

[firedog]

Yes, those can convert any rate to any other rate, so conflicting settings cannot cause HQPlayer to stop.

 

Different aspect is that when there is a change in the source format, Roon will need to add a new stream to the playlist before the currently ongoing ends. This makes the transition to happen automatically. If the playlist runs empty, HQPlayer will stop and then Roon need to ask HQPlayer to start playing again once there is more content added to the playlist.

 

I'm now playing back a mixed playlist of DSD and PCM at various sample rates. I'm getting playback stop between every track. I see in Roon that the next track cues up, but playback doesn't continue b/c of the disconnect. On some - not all - of the instances I get a quick message from Roon that the "transport has been lost" or something like that - I can't remember exactly what was written.

 

Other times I have mixed playlists that playback flawlessly. I wish I could see a pattern that would help lead to a solution....

[1markr]

I don't see it that way. I am taking advantage of all that HQP 3.12 has to offer, with the the filters and modulators working as designed, and it's working great. The fact that I can't use CUDA to offload some processing power off my CPUs is of no sonic consequence for me, so I do not agree with your statement.

 

If you do a Google search on CUDA and MacBooks, you will see many, many folks have had issues with the Nvidia drivers, and this is outside of our audio world. No more CUDA here for me, just audio bliss!

 

 

hmm if I read you well you solved issues appeared with 3.12 by loosing the benefits it brings...

[Miska]

I'm now playing back a mixed playlist of DSD and PCM at various sample rates. I'm getting playback stop between every track.

 

If you have PCM as output mode, then it doesn't work until HQPlayer 3.13. If you have SDM output mode, then it should work in most cases.

 

I see in Roon that the next track cues up, but playback doesn't continue b/c of the disconnect. On some - not all - of the instances I get a quick message from Roon that the "transport has been lost" or something like that - I can't remember exactly what was written.

 

Roon shouldn't really disconnect when HQPlayer stops due to settings conflict. Although the disconnect shouldn't cause harm as resolving the conflicting config at HQPlayer side and restarting the playback at Roon side should work.

 

I've been trying to think about elegant solution how to automatically resolve cases where settings conflict is encountered. But I don't have a definite answer yet.

[Le Concombre Masqué]

closed form is the one filter that works perfectly with 3.12 here

My Mac Mini is OK on Yosemite, mostly, using all modulators / filters to 128. Late 2012 i7 quad, 512 GB SSD, . Very lightly optimized OS. HQP runs 44.1 aiff files in Poly variants to 128 at around 1 core and overall CPU user load is ~15%.

 

I have one issue that I think is caused by the Mini and one issue with Closed Form. The Mini caused issue is a stutter and then drop out of the USB (direct connected to DAC) for a second or two then everything resumes OK. Only happens once per listening session and it can happen in the first minute or many tracks later.

 

The issue with closed form is going to 128 where at always about 2 minutes into a track the CPU use climbs to over 300% and then chokes! Going to 64 seems to work for a long time.

[Le Concombre Masqué]

Audio bliss here with closed form CUDA enabled ; can not dig why the heavy load filter works perfectly while the used to be gentler on cpu load does not work... Though I'm thankful and admirative of Miska I can not dig how this could be a Nvidia driver issue rather than a 3.12 issue. Furthermore, then, what advantage 3.12 offers you that 3.11 did not?

I don't see it that way. I am taking advantage of all that HQP 3.12 has to offer, with the the filters and modulators working as designed, and it's working great. The fact that I can't use CUDA to offload some processing power off my CPUs is of no sonic consequence for me, so I do not agree with your statement.

 

If you do a Google search on CUDA and MacBooks, you will see many, many folks have had issues with the Nvidia drivers, and this is outside of our audio world. No more CUDA here for me, just audio bliss!

[firedog]

If you have PCM as output mode, then it doesn't work until HQPlayer 3.13. If you have SDM output mode, then it should work in most cases.

 

 

 

Roon shouldn't really disconnect when HQPlayer stops due to settings conflict. Although the disconnect shouldn't cause harm as resolving the conflicting config at HQPlayer side and restarting the playback at Roon side should work.

 

I've been trying to think about elegant solution how to automatically resolve cases where settings conflict is encountered. But I don't have a definite answer yet.

 

I am using upsampling to DSD, and poly-sinc-shrt-mp filters, so there shouldn't be a conflict. BTW, the message I get is "transport lost control of endpoint".

[1markr]

Easy! I can now run Roon (thru HQP 3.12), and I can also use "closed form" now, whereas I could not in 3.11.

 

 

 

Audio bliss here with closed form CUDA enabled ; can not dig why the heavy load filter works perfectly while the used to be gentler on cpu load does not work... Though I'm thankful and admirative of Miska I can not dig how this could be a Nvidia driver issue rather than a 3.12 issue. Furthermore, then, what advantage 3.12 offers you that 3.11 did not?