HQPlayer Linux Desktop and HQplayer embedded

[Miska]

8 hours ago, craighartley said:

Sorry for another naive question, but if I change hostname from ‘hqplayer’ does that also change the ‘Username’ (also ‘hqplayer’) in Web Browser? And if not, is there also a way to change Username? I’m talking specifically about using HQPlayer OS boot image. 

 

No, it doesn't. Username and password are separate. You can change username + password with "hqplayerd -s username password". These are not specific to the HQPlayer OS image, but apply to any similar setup.

 

[Miska]

5 hours ago, Le Concombre Masqué said:

4.12.0 does not work for me in the long run (starts OK but crashes as soon as I try to compare filters) ; I'm back to 4.11.2

 

You don't need to go to the /config page in order to compare filters. The problem you describe with Audirvana applies only if you touch something in the configuration.

 

5 hours ago, Le Concombre Masqué said:

I described scenario in full length yesterday ; you answered faster that light, and this time it might not be nice for the testing might not have been thorough : you've been prompt in blaming Audirvana but Audirvana works great with 4.11.2.

4.12.X has broken something you can fix not Audirvana. And even if there's a fault in Audirvana, you had implemented a trick in 4.11.2 to circumvent it that you might have forgotten to do in 4.12.X

Plus, as I've already stressed, when Audirvana loses pairing to Embedded, the Web interface stops working as well, it behaves just as if the ip address was not correct any longer. But it is and if I reboot, I use the same ip address

 

I have not been able to reproduce any crash or that web interface would stop working, tested on NUC7i7BNH.

 

I can see that after configuration change and subsequent restart, HQPlayer disappears from Audirvana. But comes back if I restart Audirvana (without touching HQPlayer). I do most UPnP testing with mconnect Player and there I just go to Play To tab, click rescan button and reselect the HQPlayer renderer after such restart.

 

Of course, changing filters or such doesn't require configuration changes and thus neither application notice anything when I do that.

 

If you use 4.12.0, then it may cause NAA to exit with error and thus make the system non-functional. That was fixed in 4.12.1.

 

[craighartley]

55 minutes ago, Miska said:

 

No, it doesn't. Username and password are separate. You can change username + password with "hqplayerd -s username password". These are not specific to the HQPlayer OS image, but apply to any similar setup.

 

Thanks. I keep assuming booting from the HQPlayer image USB rules out making the changes possible when HQPlayer Embedded is installed on a hard disk, because I assume changes won’t stick when the system is next booted ( ie from the USB).
 

It’s my fault for not knowing how boot images load and work, but are changes (hostname, username, disabling naa or HQPlayer function, etc) written to the USB (or SD card, or whatever one is using for the boot image)?

[Miska]

3 hours ago, craighartley said:

Thanks. I keep assuming booting from the HQPlayer image USB rules out making the changes possible when HQPlayer Embedded is installed on a hard disk, because I assume changes won’t stick when the system is next booted ( ie from the USB).

 

HQPlayer image is normal in that respect, so you can make all kinds of changes that stick across reboots.

 

They just don't stick across reflashing the media (for example due to new release). Backup/restore feature resolves this, but it doesn't include OS changes. Which can be good at times too. However, you can manually backup your OS changes from command line if you want to.

 

3 hours ago, craighartley said:

It’s my fault for not knowing how boot images load and work, but are changes (hostname, username, disabling naa or HQPlayer function, etc) written to the USB (or SD card, or whatever one is using for the boot image)?

 

Yes they are, the root filesystem is always mounted in normal read/write mode.

 

 

Due to demands, NAA image is different and there any change is gone during next reboot. There the root filesystem is compressed and accessed read-only during bootup and then decompressed into RAM. Allowing you to even remove the boot media after bootup without problems (this cannot be done with HQPlayer image). For this reason, the HQPlayer image is dual-use, allowing you to use it for NAA too and then you can change the name in a persistent way, etc. NAA image can be safely shut down by just pulling the plug, while HQPlayer image should be shut down properly (usually this would be initiated by short power button press).

 

If you have only one instance of HQPlayer image running, you don't need to touch the OS settings. The NAA image being the way it is, you can practically have only one instance of it running in general.

 

HQPlayer + NAA image combination doesn't conflict because they default to different names.

 

[Le Concombre Masqué]

4 hours ago, Miska said:

 

You don't need to go to the /config page in order to compare filters. The problem you describe with Audirvana applies only if you touch something in the configuration.

 

 

 

of course I don't...

I enjoyed very much 4.11.2 with 3.6.0 and Audirvana today ; will do so till I'm ready to lose more time and/or you produce 4.12.2...

[craighartley]

1 hour ago, Miska said:

 

HQPlayer image is normal in that respect, so you can make all kinds of changes that stick across reboots.

 

They just don't stick across reflashing the media (for example due to new release). Backup/restore feature resolves this, but it doesn't include OS changes. Which can be good at times too. However, you can manually backup your OS changes from command line if you want to.

 

 

Yes they are, the root filesystem is always mounted in normal read/write mode.

 

 

Due to demands, NAA image is different and there any change is gone during next reboot. There the root filesystem is compressed and accessed read-only during bootup and then decompressed into RAM. Allowing you to even remove the boot media after bootup without problems (this cannot be done with HQPlayer image). For this reason, the HQPlayer image is dual-use, allowing you to use it for NAA too and then you can change the name in a persistent way, etc. NAA image can be safely shut down by just pulling the plug, while HQPlayer image should be shut down properly (usually this would be initiated by short power button press).

 

If you have only one instance of HQPlayer image running, you don't need to touch the OS settings. The NAA image being the way it is, you can practically have only one instance of it running in general.

 

HQPlayer + NAA image combination doesn't conflict because they default to different names.

 

That’s very helpful and instructive. Thank you very much. It also means that using the bootable image is even more convenient than I thought!

[Le Concombre Masqué]

could use 4.12.1

 

changing filters via Client

 

dealing with occasional strange behaviour from Audirvana while it worked just fine with 4.11.2. (relaunching from bottom line is a no go, relaunching from the track list is safer but still I sometimes I got the speaker or pause logo instead of the Play arrow though, of course I pressed stop before turning to Client for filter change ;   had to fiddle, even to restart sometimes)

 

Anyway gonna use long lp 2s for everything for a while I think, works with everything here (output SDM 128 convo ASDM5EC), MCH and 192 sources included. Your most resolving filter so far. Maybe mqa lp was a comfier dark with Coltrane's Blue World 24/192 Mono but the double bass was much more delineated and separated from the piano and drums with long lp 2S. Beat ext2 and al with 24/96 remaster of Kempff's Beethoven's piano sonatas : 60's tape noise and piano better separated , firmer piano. idem with recent Mozart's trios led by Barenboim in 24/48 : better spatial cues of the instruments + I perceived what sounded fuller with ext2 as boomier, revealing shrieker stuff in contrast (vastly exaggerated comments, noticeable (if not imagined !) only in a comparison.

 

backup/restore functionalities I had not taken notice before convinced me it was not going to be that much of a ordeal. However, I had to recreate the links/reload convolution filters. First launch was painful : no eQ + several dB too loud though on screen it said it was set....

 

I guess you're exhausted and with obvious priorities such as desktop but there's definitely something less smooth when using Audirvana though it seems I can live with those glitches and maybe never suffer from them again if I don't launch 10 times the same track to do comparisons...

 

overlap save for matrix would be appreciated

 

the possibility to check enable matrix/disable convo on the same page (as well as the possibility to check enable convo/disable matrix on the relevant page)  would be comfy

 

 

[Miska]

3 minutes ago, Le Concombre Masqué said:

changing filters via Client

 

Is exactly the same as changing those from the front web page too...

 

4 minutes ago, Le Concombre Masqué said:

I guess you're exhausted and with obvious priorities such as desktop but there's definitely something less smooth when using Audirvana though it seems I can live with those glitches and maybe never suffer from them again if I don't launch 10 times the same track to do comparisons...

 

It is not about priorities, it is just that I can't do anything about Audirvana. It is just another UPnP Control Point. During testing I relaunched same tracks hundreds of times from mconnect Player with lot of HQPlayer restarts.

 

But in generic terms, do not touch playback controls from anywhere else except UPnP Control Point, especially when the same instance is also Media Server. For example you cannot restart the playback from HQPlayer side, because the media URI very likely becomes invalid right after the playback has been stopped (this has nothing to do with HQPlayer). Especially for the cases like Audirvana when the Media Server is doing transcoding (if you play FLAC from Audirvana, it'll send WAV).

 

[Le Concombre Masqué]

9 minutes ago, Miska said:

Especially for the cases like Audirvana when the Media Server is doing transcoding (if you play FLAC from Audirvana, it'll send WAV).

 

Would you recommend converting to wav offline and rip + download wav to get better SQ and/or smoother operation  when using Audirvana as front end/library management ?

[Miska]

6 minutes ago, Le Concombre Masqué said:

Would you recommend converting to wav offline and rip + download wav to get better SQ and/or smoother operation  when using Audirvana as front end/library management ?

 

Only thing it would probably change is to increase disk space consumption. I'm pretty sure Audirvana sends the data always same way regardless of the source file.

 

Only non-smooth part I've experienced so far is that it doesn't rediscover HQPlayer after HQPlayer restart and needs to be restarted to get HQPlayer back.

 

[cat6man]

On 10/16/2019 at 3:19 AM, Miska said:

 

Only thing it would probably change is to increase disk space consumption. I'm pretty sure Audirvana sends the data always same way regardless of the source file.

 

Only non-smooth part I've experienced so far is that it doesn't rediscover HQPlayer after HQPlayer restart and needs to be restarted to get HQPlayer back.

 

 

i don't want to disagree but i must.

in my system (hqplayer embedded, no filter/dither/dsd/etc to NAA on opticalRendu) i hear a sq improvement with wav vs. flac.

i have no idea why, and i hadn't checked this in over a year, but yesterday i came to the same conclusion.

 

the engineering half of my brain is distressed (and curious), but the music loving half of my brain has no doubt.

 

try it for yourself..............ymmv

[Miska]

1 hour ago, cat6man said:

i don't want to disagree but i must.

in my system (hqplayer embedded, no filter/dither/dsd/etc to NAA on opticalRendu) i hear a sq improvement with wav vs. flac.

i have no idea why, and i hadn't checked this in over a year, but yesterday i came to the same conclusion.

 

From Audirvana, or HQPlayer Embedded playing the files?

 

In former case Audirvana decodes FLAC or WAV into WAV stream - in both cases it looks the same to HQPlayer. While if you use some other UPnP Media Server and Control Point to play the same files, they usually end up as-is to HQPlayer.

 

In latter case HQPlayer decodes the actual file. If it makes difference in this case, next question is what kind of media is used to store the files? Advantage of FLAC is that there's about 50% less data to transfer from the storage, while advantage of WAV is that there's less processing to do.

 

[randytsuch]

27 minutes ago, Miska said:

 

From Audirvana, or HQPlayer Embedded playing the files?

 

In former case Audirvana decodes FLAC or WAV into WAV stream - in both cases it looks the same to HQPlayer. While if you use some other UPnP Media Server and Control Point to play the same files, they usually end up as-is to HQPlayer.

 

In latter case HQPlayer decodes the actual file. If it makes difference in this case, next question is what kind of media is used to store the files? Advantage of FLAC is that there's about 50% less data to transfer from the storage, while advantage of WAV is that there's less processing to do.

 

I have a question related to this, kind of.

 

Can you confirm the mconnect (running from IOS but I would not expect it matters) is not actually playing anything?  I think it points to a file, and then hqp fetches the file and plays it.  I think something similar happens for tidal, except its more complicated because the fetching is done over the internet.  

 

At least, that is my simplistic view of things, hopefully they are not far off of reality.

 

Randy

[Le Concombre Masqué]

7 hours ago, cat6man said:

 

i don't want to disagree but i must.

in my system (hqplayer embedded, no filter/dither/dsd/etc to NAA on opticalRendu) i hear a sq improvement with wav vs. flac.

i have no idea why, and i hadn't checked this in over a year, but yesterday i came to the same conclusion.

 

the engineering half of my brain is distressed (and curious), but the music loving half of my brain has no doubt.

 

try it for yourself..............ymmv

@Superdad  criticised the Audirvana option for he perceived Audirvana has adding "a flavour" . 

 

On one hand, we now know there's an actual transformation of the files (unless they are stocked as wav), on the other hand, the processing to wav being made on the Audirvana machine, no extra processing is done by the Embedded machine. 

 

Another difference comes from Audirvana making RAM as the source

 

I hang my head : When I compared wav to flac on the same machine, I attributed beefier character to wav and more airy to flac ; when I compared origins of files, I did not like much RAM, "thinning" the music, disembodying* it, preferring wifi streaming from a local HDD (or SD card but so much less practical) 

 

Bottom line : Though @Superdad got be puzzled for a while, Audirvana, on an old Macbook Air, fronting Embedded is no handy dandy that I can't think of a simpler more convenient system (files are still wifi streamed from external HD) and so I decided that whatever flavour, if any*, Audirvana is bringing, I'm OK with the solution anyway.

 

Since another post mentions Tidal, I also hang my head there : Qobuz does not sound as good as my own rips whenever I compared (but never can be sure it's the same mastering etc), so some would say there's something wrong in my system. But maybe it's the other way round !

 

*I'm perfectly conscious that using the term "disembodying" (ever so slightly)  for RAM based playing suggests (but does not prove or state) the perception of differences as mentally projected in this instance

[Miska]

8 hours ago, randytsuch said:

Can you confirm the mconnect (running from IOS but I would not expect it matters) is not actually playing anything?  I think it points to a file, and then hqp fetches the file and plays it.  I think something similar happens for tidal, except its more complicated because the fetching is done over the internet.  

 

Yes, that's the case, so it is operates as UPnP Control Point should. Although with UPnP, there's no "file", there's a content URI that is streamed using HTTP protocol. So HQPlayer doesn't fetch anything, but instead streams it. For something like internet radio, content size is 0 meaning it is an endless stream. For a content backed by a real file, content size may have a value (known length).

 

Then rest depends on UPnP Media Server, whether it performs transcoding or not. UPnP spec places this burden and decision on the Media Server component.

 

[Miska]

2 hours ago, Le Concombre Masqué said:

On one hand, we now know there's an actual transformation of the files (unless they are stocked as wav), on the other hand, the processing to wav being made on the Audirvana machine, no extra processing is done by the Embedded machine. 

 

There's a "transformation" also if your file is a WAV or AIFF, the total transformation process is just different. Definition of "extra processing" is not very clear unless the cases are well defined (depends on network stream or local storage and in case of local storage type of the local storage used, etc).

 

2 hours ago, Le Concombre Masqué said:

Another difference comes from Audirvana making RAM as the source

 

You always have the file on disk and it always ends up in RAM? So that alone doesn't explain differences.

 

[randytsuch]

5 hours ago, Miska said:

 

Yes, that's the case, so it is operates as UPnP Control Point should. Although with UPnP, there's no "file", there's a content URI that is streamed using HTTP protocol. So HQPlayer doesn't fetch anything, but instead streams it. For something like internet radio, content size is 0 meaning it is an endless stream. For a content backed by a real file, content size may have a value (known length).

 

Then rest depends on UPnP Media Server, whether it performs transcoding or not. UPnP spec places this burden and decision on the Media Server component.

 

Thanks

OK, so I'm running embedded (loaded into Ubuntu buster) and then using mconnect to play music.  

Is Rygel the media server?  Would Rygel transcode?

 

Randy

[Miska]

1 hour ago, randytsuch said:

OK, so I'm running embedded (loaded into Ubuntu buster) and then using mconnect to play music.  

Is Rygel the media server?  Would Rygel transcode?

 

Rygel can be configured to be simple media server and it is also capable of transcoding. Since HQPlayer also uses Rygel, to have both in the same machine, hqplayerd still supports MPRIS interface with Rygel. It can be tricky to get up and running together though, so it is simpler if Rygel is running alone in a separate machine as media server. It didn't understand about DSF/DSDIFF files last time I checked though. MinimServer is another similar and supports DSF/DSDIFF, but being written in Java it can be annoyingly sluggish at times.

 

Many NAS boxes also have a built-in media server, and commonly quite simple one which shouldn't do much transcoding. It is just hard to generalize across different brands.

 

 

Edit: It may actually support serving DSF files, since there seems to be DSF plugin through libav demuxer: https://gstreamer.freedesktop.org/documentation/libav/avdemux_dsf.html

 

[bodiebill]

On 10/17/2019 at 11:41 PM, cat6man said:

 

i don't want to disagree but i must.

in my system (hqplayer embedded, no filter/dither/dsd/etc to NAA on opticalRendu) i hear a sq improvement with wav vs. flac.

 

Same here. But I see no need to store everything as wav. From JRiver I convert an album from flac to wav in 10 seconds and it is automatically put on a RAM drive and sent from Audirvana to the server PC with HQPlayer embedded and then to the endpoint PC with NAA.

[bodiebill]

On 9/21/2019 at 10:19 AM, bodiebill said:

About the PSU: currently this is not linear, but a high quality and pretty low-ripple Seasonic, for which I used the stock power cable. I made this choice assuming that the HQPlayer server is well isolated from the endpoint PC which also has a fiber NIC. I could upgrade the power cable of course.

Too early to review the sound, I think it needs some time as this is a newly built PC. If I am not totally happy I could replace the Seasonic by this combination:

PSU DC-ATX    HDPlex 400W DCATX
PSU AC-DC    HDPlex 200W LPSU (or HDPlex 100W for mobo + HDPlex 200W for cpu)

 

Any advise here on how important the PSU is for the server ('control pc') in such a setup?

Or would using the Seasonic for the mobo and a LPSU for the CPU be a good idea?

 

Just received an HDPlex 400W DC-ATX, so I compared this (combined with the HDPlex 200W LPSU) with the fanless Seasonic 400W I used before. I must say that I can hardly hear the difference. The fiber isolation between the server and endpoint PC's seems to work well.

 

This server PC is dual boot: (1) Audiolinux with HQPlayer embedded and (2) Windows 10.

With the HDPlex 400W DC-ATX / HDPlex 200W LPSU combo the first seems to work fine. However when I boot Windows and run a demanding process (such as Handbrake converting a video file), the system hangs after 10 seconds or so. Maybe this PSU combo is not powerful enough for that? Not of major importance as this server was custom built for HQPlayer-e.  

 

PS: the HDPlex combo does have the advantage that the PC runs even less hot. Now converting 24/96 to DSD256 with ASDM7EC at 45-50 degrees Celsius.

[StreamFidelity]

23 minutes ago, bodiebill said:

However when I boot Windows and run a demanding process (such as Handbrake converting a video file), the system hangs after 10 seconds or so. Maybe this PSU combo is not powerful enough for that?

 

Which CPU do you have? My Intel Core i9-9900k runs with 4.2GHz on all 8 cores (virtual 16) with

HDPLEX 200W Linear Power Supply (LPS)

HDPLEX 400W HiFi DC-ATX 

very stable. There are some settings to be made in BIOS. For example, I have limited the voltage to a maximum of 1.050V.

 

[bodiebill]

8 minutes ago, StreamFidelity said:

 

Which CPU do you have? My Intel Core i9-9000k runs with 4.2GHz on all 8 cores (virtual 16) with

HDPLEX 200W Linear Power Supply (LPS)

HDPLEX 400W HiFi DC-ATX 

very stable. There are some settings to be made in BIOS. For example, I have limited the voltage to a maximum of 1.050V.

 

 

 

i7-9700k.

 

Thanks for the suggestion! I will have a look at the BIOS and see whether I can make Windows 10 work under a heavy load with maximized voltage...

[bodiebill]

24 minutes ago, StreamFidelity said:

There are some settings to be made in BIOS. For example, I have limited the voltage to a maximum of 1.050V.

 

Which voltage? CPU, DRAM? There are many options in my ASRock BIOS.

[StreamFidelity]

55 minutes ago, bodiebill said:

Which voltage? CPU, DRAM? There are many options in my ASRock BIOS.

 

Unfortunately, I am not a BIOS specialist. I oriented myself to Gaming Tester on Youtube. The picture shows my CPU Core Voltage setting:

 

- Adaptive
- OC Voltage 1.100V
- Offset Voltage -0.050V
= 1,050V

 

Be careful! Always save the working setting under profiles! I would still install the latest BIOS. Before that, be sure to back up the settings to USB Stick. In my BIOS, everything was overwritten with update.

 

[rickca]

1 hour ago, bodiebill said:

However when I boot Windows and run a demanding process (such as Handbrake converting a video file), the system hangs after 10 seconds or so.

Your i7-9700K is base frequency/turbo frequency of 3.6/4.9 GHz.  It can sustain an all-core turbo of 4.6 GHz.  Handbrake is most likely running multiple cores at high turbo frequencies.  Without some kind of constraints, there's no doubt you will exceed what the 200W LPS can deliver.  The 95W TDP is what's required to run 8 cores at the base frequency of 3.6 GHz.  You can easily double that at all-core turbo.

 

Your motherboard probably isn't contributing much to the load since you don't have a video card, so maybe something like 35W.  Even if you power the motherboard with your Seasonic, and just use the HDPLEX gear to power the CPU, I think you may run into the same problem.