Euphony OS w/Stylus player setup and issues thread

[Johnseye]

2 minutes ago, dminches said:

 

Does the “N” show up on your screen but get ignored?

 

 

No.  After the text says to select Y/n it always defaults to Y without any "n" appearing on the screen.  I have tried it with 2 different keyboards.  Euphony does recognize the keystroke at the entry screen where you get the option to select Euphony, Reboot or Poweroff, so I know the keyboards are working through the BIOS at that point.

 

[Johnseye]

1 minute ago, dminches said:

 

I would open a support ticket.  He responds very quickly.

 

 

Yup, already done. Thanks.

[Johnseye]

For anyone having the same ramroot issue I had, Željko remoted into my PC after booting from a USB stick and corrected it. 

 

One thing to note is that I had copied Euphony directly to the hard drive instead of installing it from the USB drive. Not sure if that contributed to the problem. 

[Johnseye]

11 hours ago, davide256 said:

Try using a wired vs wireless keyboard.

 

Exactly what I did. One of each

[Johnseye]

9 hours ago, edwardsean said:

20210216
- tags that are not directly indexed by music DB are accumulated in "otherfields" tag - which is now used in search and in filters
- add menu items for switching to another Euphony and shutting down both machines only if other machine responds to network query 
- added more checks to registration dialog (to make clear when people confuse voucher code with registration code)
- clarified some messaging in Complete Installation dialog
- Roon core updated to 1.8
- improved internal/external drive mounting and detection
- reconnecting network interface after Play and disconnect now possible even for WiFi interface
- added IRQ handling in Expert Settings CPU isolation
- fixed UPnP for some UPnP endpoints (RoPieee and SOtM confirmed)
- added UPnP conversion option: bit-depth 
- fixed supported rates detection
- fixed "Song is not accessible on path.. " error 
- fixed Qobuz playlist content limit of 50 items
- fixed problems that appear when you fill the RAM with buffered songs from Queue

 

Excellent. This is a big update.

[Johnseye]

16 minutes ago, mrkoven said:

Anyone switch from Roon+HQP to Euphony+HQP and can comment on the experience in SQ difference, if any?

 

Yes, I did that last night.  I've been listening with Euphony+HQP for a while.  I switched to Roon+HQP last night only for a short time.  To my ears Euphony continues to sound better.  Clear, more defined instrument and vocal separation.  Will be putting more time into comparing over the next week or so before settling on one.  I really wish the Roon+HQP could improve to the Euphony level, or the Euphony interface would improve.

[Johnseye]

16 minutes ago, jbm_2021 said:

I would like to give Euphony a try. Is there any device that can run Euphony server with an external clock capability?

In other words, has anyone tried a Fanless NUC with external clock or something similar?

 

 

9 minutes ago, mrkoven said:

 

Thanks yeah it's tough to beat Roon UI/UX but ultimately if SQ is better than it would be worth the switch for me.

 

What are the details of your custom audiolinux server running Euphony+HQP?  Is it an Intel NUC?

 

My "server" is a mini iTX board that had all its clocks modified to tap off an sCLK-EX that also supports a tX-USB ultra and tX-USBexp.  It's running Euphony with HQP Embedded now, although I had been running Audiolinux on it.  I also have an Intel NUC that runs Audiolinux in NAA mode.   I pulled the board out of that NUC and put in its own fanless case.  The clocks on that NUC board were modified to tap off another sCLK-EX that also supports a sNH-10G.  The Mutec Ref10 provides the master clock for those two sCLK-EX's so in turn the server, endpoint NUC, tX-USBultra and switch.

 

I could run the HQP NAA as a stand alone image or Euphony running the NAA.  I like some of the tweaking I can do with the Audiolinux and in the end it's only running the NAA service.  I can load the entire thing to RAM and pull out the USB drive which I like as well.

 

The NUC could run Euphony or Audiolinux on its own but if you want to upsample DSD to 512 I don't think it's well suited.  I do upsample to 512 with the iTX server which is why I use the NUC as an endpoint.  I wanted to isolate any noise from the heavy upsampling.

 

The NUC model is NUC7I7DNK1E.  I have another unmodified one I use as an endpoint in a different room as well.  It doesn't get any use so if anyone wants to buy it let me know.

 

[Johnseye]

13 hours ago, jbm_2021 said:

Thank you John. In your post in this thread, you shed some more light 

What exactly are the mods needed to a motherboard and which motherboard exactly are you using? If I buy a fanless NUC like this one https://www.onlogic.com/computers/nuc/?cpsrc=Search_Alpha_A&kw=fanless nuc&gclid=Cj0KCQiAvbiBBhD-ARIsAGM48bzjMIlvs2ESx8rtGnImdFiw3LH-eVskMcfjP6j6IS0GZQ4HeQEds5QaAjCjEALw_wcB

will I be able to still use the sCLK-EX to achieve the clocking? Please let me know what do I need to look for in the mobo for this mod?

Thanks a lot!

 

 

There are a few clocks used by every motherboard that directly impact SQ.  The processor clock is the most critical.  The other two are the NIC clock and USB clock.  SOtM will tell you that modifying every clock on the board will have an impact.  The three I identified are the most impactful.

 

The clocks on the board will need a connector soldered to them.  A detachable cable connects the mobo clock to the sCLK-EX. I had SOtM do all the soldering work.  I shipped my mobo to them in Korea.  Also buying the sCLK-EX directly from them they matched the frequencies and tested the clocks with the board.

 

Any mobo can be modified.

[Johnseye]

Does anyone have difficulty with Stylus either not playing song after song consecutively in the queue or just have a general difficulty playing songs at all?

 

I use Stylus with HQP Embedded.  I buffer each song 100%.  I almost never get a consistent listening experience.  Always struggling to get songs to play.  Not sure if it's Stylus, HQP or both.

[Johnseye]

6 hours ago, Joerg D said:

 

Hi,
it may be because of the Hqplayer. It needs a lot of computing power. I had dropouts. Without a Hqplayer, the music plays immediately. Temp / CPU then shows a red bar.

Jörg

 

 

It may be because of the sync between Symphony, HQP and the NAA.  It isn't a hardware issue.

 

[Johnseye]

Not long ago I posted about an issue I was having with songs not automatically playing from my queue with Stylus.   I received a response from Željko about it.  Indeed it is a bug and a bigger issue than expected.  I also noticed HQPe was not recognizing my tracks as 24 bit.  It thinks everything is 16 bit.  Both of these issues are tied to the same Stylus bug.  Here is Željko's response.

 

Even though I really like how Stylus sounds, I'm going to start going in another direction until this is resolved.

 

"this is an issue with the current limitation of Stylus - HQPlayer integration.

It was initially conceived that since HQPlayer plays only limited set of song formats if HQPlayer-incompatible song comes for playback Stylus would take over and play the file.
So everything was organized around Stylus being able to play a song if needed. In your case, where there are no local audio outputs, Stylus player logic is in a weird disabled state. While it can still set a song for HQPlayer playback on explicit command it does not properly take care of the song transition situation.

I'm just explaining why it doesn't work, not justifying - this is obviously a bug - as we added the ability for HQPlayer to play on remote NAA where local audio devices are not needed we did not properly adjust the remaining logic for playback continuity.

If you had internal audio cards you could enable them so Stylus could at least think it can do playback but since you do not have any audio device whatsoever there is no solution until we fix this.
If the solution were simple I would have fixed it immediately for your installation but I'm not sure what exactly I must do and how many changes would this require so you will have to wait for the next release.

 

Both issues are connected to this same bug - even though HQPlayer is actually playing a song, Stylus player logic does not know it, does not check for playback status, position update or input format and samplerate..."

Sorry for the inconvenience."

 

 

[Johnseye]

Update: Željko took a further look and all was not as it appeared.  He found another cause and implemented a workaround.  I'll just quote his response here:

 

"In your case HQPlayer does not correctly report the song position so the position update from HQPlayer just stops 5-6 seconds before the end of the song (I guess in reality HQPlayer finished playback already?!).
Since logic relies on song position and expects that playback will last at least for the song duration +-N seconds and since N was set to 3 - in your case the song end was not picked up and End-Of-Song event was not triggered (so the next song was not played).
I just set N to 7 on your installation and everything works (except you will have 5-6 seconds pause between songs - which I cannot do anything about - I cannot guess when will HQPlayer actually finish playback).

There were some errors in the log related to the bug I described earlier but I don't think this bug was part of your issue.

Input samplerate was another issue (not exactly a bug but it needs to be fixed via the correct settings presentation).
I have to check again but last time I tried HQPlayer did not want to play downloaded Qobuz songs. Because of that, songs are automatically converted to wav format after downloading and this conversion uses "UPnP conv. bit depth" settings. Default was 16 so I set it to 24 now. I will have to make clear in the next version that those conversion settings are applicable for UPnP and HQPlayer playback both."

[Johnseye]

6 hours ago, stefano_mbp said:

It seems the issue is all on HQPlayerd (Embedded) side as there are other issues with upnp gapless.

Unfortunately, Jussi doesn't see gapless with upnp as a major problem ....

 

If it were purely an issue with HQPe we would also see it with Roon + HQPe but that's not the case.

As far as the gapless specific issue is concerned, this isn't a problem for standard albums, but for recorded concerts where the entire show should be seamless gaps are a big issue.  It destroys the continuity of the listening experience.

[Johnseye]

1 minute ago, stefano_mbp said:

It is an upnp related issue and Roon don’t use upnp

 

I wasn't using upnp before but Željko enabled it to get the queue flowing automatically.  Ideally upnp isn't necessary in the first place and files don't need to be converted to WAVs for playback.

 

Also to note that the queue stopped automatically playing songs again last night so I don't think my issue is resolved.  Not to mention that all tracks are now being viewed and reported as 24bit instead of them all being reported as 16bit.  I don't know how this impacts HQPe's processing of those songs and whether it impacts SQ or not.  It's concerning regardless.

[Johnseye]

7 minutes ago, BCRich said:

I was specifically told that gapless playback via HQPe in Euphony/Stylus is not presently supported.

 

That is part of the issue at hand. 

 

[Johnseye]

My understanding is that Željko is talking with Jussi.  Hopefully the issue regarding a silence gap between tracks gets resolved soon.

 

Željko has again resolved the issue I was having with songs in the queue not automatically playing.  I am back to not using upnp.

He has also resolved the issue I was having with the track bit depth not being correctly recognized by HQPe.

 

I am very happy about those issues being resolved and am hopeful of a gapless playback solution.

[Johnseye]

1 hour ago, dminches said:

 

To be fair, Stylus has no issues playing gapless on its own.  Asking each software vendor to keep up with other software packages is asking a lot.  Every time there is change to HQPlayer Željko has to make sure it still works with his software.  I am sure he will get there but it is a burden.

 

 

I believe he has this working on my system.  I did verify that it works, but have another issue, so almost there.

 

Željko or any software vendor doesn't have to do anything they don't want.  We as customers would like this functionality, so if he wants to provide it, his existing customers will be happier and stay with his product.  He should also gain new customers by including this functionality.

 

I think when you decide to incorporate another piece of software with your own, you make the decision then to include all of the expected functionality of that other software.  You may rush to market and not have it done at first release, but I would expect the intention to provide it.  Željko has spent some time on this the past day or two and appears to have done it.  Not bad from a time spent perspective.  Complaints will be down, customer satisfaction improved and he may gain new customers as a result.

[Johnseye]

Some good news to report.  Gapless playback is now working as intended with HQPe.  Željko, Jussi and I worked through all the issues.  Željko has stated if anyone else would like these changes made, and be considered a beta tester along with me, to let him know.

 

Some things I learned along the way are:

 

In Stylus, "Convert non-DSD formats" doesn't need to be selected for PCM.  The wording of that selection is misleading. This selection causes all songs to be converted to wav before playback and conversion bit-depth is fixed (does not take into account the original bit-depth of the song).  It was causing me a bit depth issue.  So it's not necessary for PCM listening and you shouldn't need to check it unless I guess you're listening to mp3 in which case you need to do some soul searching.

 

In HQP, forcing a sample rate while having "auto rate family" checked can cause issues.  Per Jussi: If you set "Sample rate" it is explicit output rate request. If set to "Auto", highest suitable at or below limit is picked up. Using "Auto rate family" is in direct conflict with setting "Sample rate" to anything else than "Auto".

[Johnseye]

1 hour ago, bos3812 said:

1A is fine.

 

I've had issues with the USB XE and an Uptone LPS-1.2 @ 1.1A.  I would recommend more than 1A.  Since using a PHD SR4T @ 2.5A I've had no issues.

[Johnseye]

Does anyone have experience with importing an .m3u playlist into Stylus?  I can create a playlist natively within Stylus but am having a challenge getting it to use an imported one.

[Johnseye]

13 minutes ago, TheAttorney said:

I've had difficulty trying to understand what problem you're trying to resolve.

 

I think what you're effectively saying is that, when buffering files into RAM, noise from the network somehow enters and stays inside the RAM alongside the 1's and 0's - thereby permanently corrupting the file in some way. And you can hear the resulting SQ degradation when you later play the buffered file - compared to when you had buffered the files into RAM when the network was disconnected.

 

How did you run this test to compare the SQ of the two different ways of buffering the file?

I don't see how you can achieve this test without Z creating the new feature you're asking for.

 

network activity during playback can negatively impact SQ.  disabling the network during playback stops that activity.

this can be tested by disconnecting a network cable or disabling the nic.

[Johnseye]

46 minutes ago, di-fi said:

Somewhere above davide256 summed it up very clear to me, what happened to the song before you hear it. (italics are my comments)

 

quote:

 

'a) caches the queue (to your SSD or Nvme or Optane wherever Euphony resides)

b) takes network connection down

c) buffers queue from cache (to RAM buffer from the cache)

d) plays entire queue after buffer completes

e) brings network connection up after play completes or upon any error

 

end quote

 

So I would think in a) there would be network pollution added to the cached file, with network connection still on. Why would it 'filter out" before the buffer state in c)? Please help me understand.

 

It isn't about the data.  It's not the song cached in RAM.  It's the noise generated by the NIC during playback.  Unless you can validate the NIC is not utilized at all, which is what bringing it down does.  If the NIC is utilized for anything during listening it can generate noise.

 

[Johnseye]

2 hours ago, di-fi said:

Great, I got you, I think. I will try to recap.

 

The data in the cache and the data in the buffer are identical. They both include the same good or bad stuff added. Right?

So then playing from either cache or buffer depends the quality of your memory, for example SSD compared to Apacer RAM?

Leaving the Network on will have the same negative effect on each, whether playing from cache or buffer.

 

I really appreciate your explanation Johnseye and NanoSword!

Thanks, 

Paul

 

 

The data is the data.  Nothing gets added, although some people will tell you ripping on a quiet PC sounds better than on a noisy one. YMMV.

 

There are a lot of variables that can impact SQ.  Noise is one.  Jitter is another.  Latency is another.  Reducing all of those to as little as possible has been a goal of digital music reproduction.

 

I don't know how the quality of RAM impacts SQ but probably by one of the things I mentioned.  I don't know enough about it personally.

 

If the network is on it can create unwanted noise.

[Johnseye]

1 hour ago, mrkoven said:

Is there any issue with stylus playing pggb 32fs? 

 

With or without HQP? 

 

There were some issues at first but they were resolved. Should work fine. 

[Johnseye]

Has anyone noticed any issues with HQPlayer embedded? I was playing some test files and the center imaging file only plays out of the left channel. Likewise I played a different file that is supposed to engage both speakers and only the left speaker played. I can play left or right channel files separately and they play fine.

This occurs both with our without an NAA so it is being initiated by HQPe.

I then played the same center image file through Stylus without HQPe and it worked fine.

 

I've attached the center image file.  You can also get it as well as many other test files from Free Online Audio Tests, Test Tones and Tone Generators (audiocheck.net)

 

If this is a known issue already discussed please let me know.  Otherwise, if you are willing, please play the test file with Stylus and HQPe and let me know if sound only comes out the left channel.

audiocheck.net_C.wav