XXHighEnd

Hi Peter,

I installed some days ago the OS and since then I was hearing some ticks and cracks. Today I´ve installed the new  0.9z-7-5 and two or three ticks are still there one just at the start of the first track in the Playlist and the other ones a few seconds inside this first track. No more cracks when changing volume or at stop.

Some questions,

- with this version I understand that it is not necessary to install again XXengine3.z74.rar in order to avoid the cracks when playing with low SFS, as it was only made for the 7-4 version. Correct?

- Not always but frequently after playing some tracks when I stop XXHE and turn it Off the X icon disappears from the task bar but when I click again in the desktop icon to turn it On a message appears: XXHighEnd is already running. Kill the other instance/s?... but the program is not running anymore...This issue was happening also with the previous version the same way that happens now with this new one. As it is happening after I re installed the OS maybe I did something wrong.
In the Task Manager I can see XXEngine3.exe and three or more XXHighEnd.exe depends on the times I´ve tried to turn it Off/On and I can´t kill them with End Process, to do it I have to reboot the computer.

Best regards,
Juan

Peter, I realized that it takes a few minutes to appear XX again after I clicked in the XXHighEnd is already running. Kill the other instance/s?... message. The Playlist is still there even when I cleared it before.

Juan

I set the parameters to suppres the cracks to there default value. After a stop i clearly hear one or two cracks before it stopped. So i changed the parameters to the old values.

Hi Peter,

Downloaded and installed z-7-5 then hooked up my voltmeter to measure AC pulse at speaker binding posts as usual.

At low SFS (2) and low net buffer size (15x1x4096) , z-7-5 brought back audible ticks at changing volume and powerful LF plop at stop - the same way as I experienced in pre z-7-4. Increasing SFS and net buffer yield even worse annoying tick and plop.

I will adhere to z-7-4, the version that I can tweak (my XXHE setting, bios setting, device driver choice) to get softest plop at stop (very mild LF pulse). Moreover, mild tick will occur only once at the first click of digital volume change in a particular track. The subsequent change in digital volume in that same track will produce no more tick at all.

Luckily, this low SFS and net buffer size yield a better SQ in my system too.

Kind regards,
Praphan

I realized that it happens to me as well, besides all the other things I stated in the previous post. I installed again the OS.

Best regards,
Juan

Gosh I'm reluctant to download this version also. The last one(Ver4) was a mission to get working and its certainly working fine for me now. I'm beginning to wonder if Linux is a better option perhaps Windows has had its day for audio! Really I'm over replacing the OS to recover from glitches, we just want the music. I have to say I'm on the road to a dedicated server build.

Of course not. I mean, it's just me - so to speak. Well, sort of.

I guess the first thing you guys must do is deactivating Phase Alignment. Ehm, it is too difficult ? for me the same ?
I'm almost serious because what I read so far in this topic all doesn't seem to make sense. So, that is a problem too - how do you make clear towards me what is going on.


Johan, what do you think the default values are, and at which values it did not work for you ?


Juan, you got it right. Point here is that these first ticks you may perceive are outside Arc Prediction; this is about showing the Wallpaper which is too offending for the system (and to me it seems that the faster the system the more offending it is). So, it stalls the stream briefly and this is what you now hear with Arc Prediction. Shut off Arc Predition and it will still happen, only now you must be able to perceive a few ms of silence. :yes: Anyway, this is what the "After SFS Rounds" setting is for.
Maybe I should have mentioned in the Release Notes that this really is unrelated to Arc Prediction itself (hence it can't be solved in there).


But Praphan, why ? Did I ever ask you to do this ? So ...
So you thought it would be a good think to measure an as low "LF Plop" as possible ? haha. Can't work. Can't work because too many parameters play a role at the same time, and it's even so that the "better" that LF plop is there, the better Arc Prediction works. And no, now don't try to make it as large as possible. :) :)

What I *can* try to "teach" is that, generally, the more SPL coming from stop, the more fast the "fade out" went. Now think about the relation to the volume - the higher the volume the faster fade out is needed within the given time (which is your SFS and also net buffer size).

What I sure can admit is that when things are not working properly that fade out can shake the house (say all the doors). But as far as I'm concerned this is not the case at all in any of the playback means which allow Phase Alignment. Still : the lower the SFS the more firm it will be.


My official response here should be : impossible. And following from that : you must have something else wrong. However, two things : 1. Of course if yuo say it I believe you, and/but 2. you (again) should be oberving in decent fashion.
Still I can't believe it ...


And this neither. :swoon:
Here though there might be something which wasn't thoroughly tested and this is about making changes to settings and to what degree they are processed "decently". So Praphan, it might be something for you to retest a few things, but now first kill XXEngine3.exe before you start playing with a new setting. If you also test other formats, I'd again kill XXEngine3. And a bit of an "inside" problem : when I tested everything (which took two full days) I need to kill XXEngine3 which is log file related ...

So let's hope this makes things better and then I know where to look for.
Otherwise, please try the setting I have been using without any single issue since 0.9z-7-5 (which was Sunday) : SFS=60, Q1=14, Q1F=1, DevBufSize = 4096, volume -30 to -18 and changing it all the time.

One last thing that I notice myself now (and which has been reported by others) : Most often I quit the album before it's finished. So, press Stop (or Alt-s). However, the one or two times the album finished by itself (or End of Track was engaged) I seem to notice that more "shaking" end. So, if you mean *that*, then by now I should be looking in that direction. Could also be important when you further (like to) test, and just pressing Stop each time may incur for nothing serious.

Really, I hope it hasn't gone too difficult ...
Peter

Well, if I see how many times (eh, per day sometimes) you guys reinstall the OS because you think that will help ...
I'm afraid it won't go otherwise for you, but I wish I could make clear in one post how to know whether the reinstall of the OS may help you with ... yea, what. It hasn't been the solution for me one single time ever, but that's part of your problem ... I seem to know what I'm doing. But to put it as it should be : I pay attention to everything and when suddenly something changed for the worse, I tend to know fairly quickly what did it. So, it's about the paying attention, *and* to the proper things. And this can't be explained ...

Once you have the OS running like you think it should, make an image copy of it (Acronis True Image really does a good job) and when needed you can restore that in 20 minutes or so (keep the partition of the OS small) with all your settings.


Of course this helps a lot, if not all; when you perform other normal work on the Audio PC - let alone if others are on that PC as well - then it could be too much of that payting attention for a normal human being.
Still it requires some specialities like not switching on Windows Update.

I don't think this helped a lot. I still tried ...
Peter

Peter, what could be origin of the problem I find and that I explain it in my first post:
- Not always but frequently after playing some tracks when I stop XXHE and turn it Off the X icon disappears from the task bar but when I click again in the desktop icon to turn it On a message appears: XXHighEnd is already running. Kill the other instance/s.... This issue was happening also with the previous version the same way that happens now with this new one. As it is happening after I re installed the OS maybe I did something wrong.
In the Task Manager I can see XXEngine3.exe and three or more XXHighEnd.exe depends on the times I´ve tried to turn it Off/On and I can´t kill them with End Process, to do it I have to reboot the computer.

The OS that I installed has 98 updates, with them installed the problem doesn´t exist but if I eliminate the updates (but the only one impossible to eliminate) the problem appears again, I´m trying now to know what are/is the update/s that avoid the problem. Although maybe the problem is not due to the updates but a bad XX configuration in my computer

Juan

Look Windows is an absolute pain. After reloading I have to spend time shutting down services and then I'm never sure I've got them all. Dedicated software does away with this.

Lets face it, Windows is designed for everybody but not "us dedicated audiophiles". Honestly its got to be simpler.

Perhaps you need a windows version that suits everybody and a dedicated audio version that is designed to suit exact or near minimum hardware setups such as yourself and Juanpmar.
 
In saying that I do appreciate all you have done and I do love the sound I have with my cr*ppy laptop which still beats my CD setup. PA is still a viable thing. The problem is making it work across a multitude of setups. The DC thing is a problem for most. They don't have the insight or gear to do it.

We have to try new things in the digital domain to push the boundaries and you are certainly doing that. 

This is mean't as constructive advice not a criticism, I want XXhighend to work "easily" and sound amazing. Its got to be simple to work.

I have no proof but I'm blaming Windows!!!! Its never been right!!!

Robocop ... I don't get it ...



This is what XXHighEnd is doing for you. So, what do I miss here ?
Or you maybe ? :)

Peter

Juan,


The only thing I recognize myself is that 0.9z-7 is prone to not get out of memory as fast as before. And I think I know what causes this although I never really tried because it doesn't bother really. Well, not me. But as it seems it bothers you ? ... and not sure how OS updates can be involved with this ...
Anyway for sure XXHighEnd.exe will vanish from memory but it may take 5 seconds sometimes. Not that you will see that on the GUI (which disappears right away).

Btw, if the solution is not to install those updates, then so be it. But is it ? it is not 100% clear to me.

What you can do to confirm my guesses about this, is go to teh Scaling button ([ S ]) and remove all the references to image files. So, for button backgrounds, form backgrounds, everything. I think that may help ...

And maybe superfluously : when XXHighEnd keeps on running you pretty much can't do a thing (well) anymore; mechanisms are in there that wait until the other version has closed down before proceeding (changing the Volume during playback would be an example of it).

Peter

Hi Peter,
I like to tell my experiences with 7-5. Since I use a 24bits Audiophilleo I can use PA now for the first time.
As DAC and Pre I use a NAD M51 with digital volume control. So the Digital Attenuation is fixed set to -10.5. I only have soft cracks during volume changes on the dac.
At stop I do not have a crack with Phase (-) but when I use Phase (+) I get a louder crackling sound on my right speaker at stop. So I stick to Phase (-). At stop I get a small repeat before it stops.
So far I really like the sound!
kind regards

Hi Peter,

Thanks for a thorough elaboration.

I developed a habit to measure the LF pulse since I don't like it even though it is not audible. This is because I hate to see large excursion of the woofer diaphragm. I know it is not harmful but this is some kind of "obsession". So what I am looking for is to eliminate the LF plop rather than accentuate it. This is achieved with my low SFS and net buffer setting in z-7-4. The fade out at stop is fast though but at a higher audible frequency,  very mild and felt unharmful to me.

I must admit that I did apply an undecent test on z-7-5 last night. I think what went wrong yesterday was that I tested 7-5 in full OS mode and attended.  I am quite certain that I clicked minimize OS after activating this new version which should reboot the system into MIN mode. But when I booted my audio PC today, it was not. The min OS button was blue. So I had to reboot into min mode. Therefore the 7-5 test last night was conducted under the full OS mode (attended too) which was not decent. Moreover, I used small bookshelf speaker and therefore audible crack was heard at changing digital volume. When I changed the net buffer setting and SFS size, I also gave too high the value.

I conducted a more decent test on z-7-5 again today at your recommended setting NB 14x1x4096 and SFS 60. You are right, no audible tick/crack at changing volume. But instead an inaudible mild LF plop *each and every time* the volume is changed with about 3-4 second delay after the command. Just watch the woofer diaphragm. At stop or next track command is given, larger (than that of changing volume) LF inaudible plop is seen on diaphragm. However, this is at a lesser excursion than what I saw on z7-3 and before.  I think I can leave with this if you insist that LF plop things should be allowed and maintained which indicates that AP is working at its best. Which in turn yielding a better SQ ?

So my bad and switching to 7-5 today.

Thanks and regards,
Praphan

PS : I also noticed that more complicated thing is going on at the end of the playlist when the streaming stop. Not only simple plop but combined with crackkkkinggggg but short though. So what I usually did is hit Alt-S like you, or even turn off the main amps before the end of playlist if I am done with the listening session.

- Well, it bothers because several XXHighEnd.exe are running at the same time although they can only be seen in the Task Manager. The problem is if that there is more than one XXHighEnd.exe running at the same time it is impossible to load a new Playlist and play music. As I say the only way to solve it is to restart the computer. I´ll try to see if waiting for more than 5 seconds after I turn it Off solves it.

- It is supposed that the updates don´t have to be installed in order to get the best possible SQ. With the updates installed don´t exist the problem mentioned but of course I have to try to get the best SQ. At the end I took out all the updates but the only one that can´t be uninstalled.

- I don´t have anything clicked in the Skin Settings, so plain GUI.

Btw, no more clicks since I use your settings (SFS=60, Q1=14, Q1F=1, DevBufSize = 4096, volume -30 to -18) and:

- After SFS rounds: 2
- Fade-out Step (3): 3

Juan

Haha Praphan, that really would go too far, wouldn't it ?
Anyway, I will look into this now. I mean, there must be some truth in the difference here.

Thanks,
Peter

So, try to understand this :

The change applied at Volume Change goes faster the smaller the SFS is. In this case this is NOT related to the net buffer settings. Now :

Depending on the "speed" of change needed, this will be low level frequency or higher level; when it will be too high for your Bass-Mid XOver it won't be thrown through your woofer but through the mid(/high). This is how a thump because a mere tick. And also :
Once it works wrongly (which it did u/i 0.9z-7-4) it is a high transient hence firm tick. So, it is the "art of the skill" to have it as low as possible but not too low because it would be to much "sub low" make the house shake, and don't make it too high because it will be too transient and imply a tick.

Now you will understand how the lower SFS will create higher SPL and higher frequency, although it would be a one-direction pulse only. This is also how the mere "thump" is more convenient than the high transient tick.

At stop though, all is also dependend on the net buffer size because it is -thus far- too difficult to spreak the fade-out over more than two buffers. This can be critical because at e.g. a Q1 of 1, Q1F of 1 and Device Buffer Size of, say, 512 samples, the fade-out happens in 1024 samples. Next imagine how at a speed of 705600 samples per second these 1024 samples only pass. This is FAST. But now we can see that it is also related to tyhe sampling speed. So, what is relativey easy-peasy for 24/88.2 is already 8 times more critical at 24/705.6. Add to this that the volume level itself plays a main role and ...

well, maybe we can see how difficult it is to give guidelines as I tried earlier on, but also how sheer infinite testing has to be to test it all (and not forget a few combinations).

But I sure hope it has become a bit more clear now.
And maybe it has also become clear by now what happens if only one sample is missing in the stream.
Yea ?
Well, a fade-out as we talk about it here within one sample. Plus a fade-in (when the stream continues) of again 1 sample.
And then to think that ever back we had all these problems of W7 where half of us suffered from hiccups and stuff because in W7 the priority arrangements changed. No, that would not be much suitable for what we're trying to accomplish with Arc Prediction.
So think og it again : 705600 (or 768000) samples per second which technically are 705600 x 2 x 4 single bytes = 5,644,800 bytes per second which must totally 100% keep on streaming and in line etc. and in between that something asynchronous like async USB is to operate;

Before 0.9z-7 you could say it was quite full of small glitches but nobody could notice; you won't hear one sample missing. Still, now that is out by guarantee (Phase Alignment is the proof, or otherwise you'd hear the highest transient ticks for sure) - SQ should have improved for this reason alone.

Ok, so far for now ...
Regards,
Peter

Nice to hear Arjan. Thanks for sharing.

And about the Ph+ (at Normal Phase) ... I tend to remove it. It only makes things more complicated and officially it can't work anyway. It's only that I can't reason out well whether in "circumstances" Ph+ could be beneficial at Normal Phase.

Regards,
Peter

Thanks for a clearer explanation on things going on during and the end of the streaming process with PA feature engaged.

Seems to me that PA has added more data crunching complexity to the streaming process (as opposed to non-PA streaming) so much that any interruption like changing volume or applying next/stop require meticulous handling of sample data to make sure nothing will be lost so much that glitches can be generated in an audible frequency band. The compromise is to put those glitches into LF band with minimum amplitude so we don't feel the earth quake!  Wow ! What an art of coding you have done!

Not sure what I understand above  is something in a right direction.

Thanks,
Praphan

Hi Peter,

SFS =2 gives the best SQ in my set up. However it is accompanied with some problems.

Firstly there is still a loud tick within the first 10/15 secs. Secondly after 15/20 mins of playback there are a series of ticks for about half a second. Thirdly there is a loud LF thump at clicking stop.

Best regards,

Arvind

Hi Arvind,

So these are related to PA engagement.

1. The first tick should be related to Wallpaper. You may like to try to extend the delay in engaging PA by increase the SFS round parameter.

2. I used to have a constant tick, tick , tick ,tick......... But for my case, it is track specific especially some albums copied from a friend of mine who ripped his albums into AIFF (illegal may be  ;) )  This implies glitches in the streaming process.

3. LF pulse at the end of track or changing track is inaudible for my case. But I can see a large excursion on the woofer. Harmless though. You may like to change your SFS to 2/2 to see if it is inaudible.

Praphan

What Praphan said.


Seems a typicle case of trackboundary and something being too slow. BUT, maybe your logging is on. This is sure not advisable for the lower SFS range !


Is your Q1 still at 14 (like your sug tells) ? That, with Q1F at 1 and DevBufSize of 4096 should be okay for the NOS1.
Also, it depends a bit on how you experience the "thump". So, it is very low frequency if all is right. Could even shake your doors. But with Q1 (etc.) too low, it will be more towards higher frequency up to a louder tick.
But otherwise it is completely normal or at least unavoidable at the lower SFS settings.
And of course the "Intensitity" plays a role. So, the higher that is, the more firm the effect. Volume level the same.

I hope this helps somewhat ...
Regards,
Peter

Hey Praphan, this is interesting, and should be related to my inability to remove the ticks between tracks *and* in between track parts (at SFS boundaries). So, always one of these phenomena is there and it is so for years, as we know. However, I can see on the analyser that it is in between track parts as well and in a fashion of which I thought it should be audible. But ...

But now your text 100% goes in the direction of other data in the header than what's "minimal" (could be embedded Coverart, whatever), and my test signals also contain some other data ...

Shall this finally lead to the solution of those ticks ?

Peter


PS: So, most probably there's nothing wrong with those rips, illegal or not. It is just that I (now think I) am doing something wrong with those headers, and AIFF is prone to contain more than the "minimum".

Hi Peter/Praphan,

SFS rounds is set to 9, which is the max.

Logging is switched off but the series of ticks continue.

Best regards,

Arvind

Hi Arvind,

You said "after 15/20 minutes";
Can you try to find some (classical) piece which lasts 30 minutes or longer, or anyway which can tell you whether these series of ticks indeed happen at the track boundary ?
Notice that testing with 10 minutes or so may not incur for "proper results". So, longer than your said 20 minutes will. Thus, *or* the ticks now will happen after this longer track has been played (a next one coming behind it), or it's still at your 15/20 minutes. The result may tell me something.

Important : Do notice that you are not in Minimized OS Mode (derived form the other topic), so I am not sure whether this all isn't a moot thing in the first place (read : better solve that first !).

Regards,
Peter

Hi Peter,

After min OS the situation has changed significantly. I have changed to 60 SFS with SFS round =1. Now I get one tick within a min at the start of the first track, none after that. The series of ticks after 15/20 mins has disappeared. At stop playback there is a LF thump but the intensity is much lower than earlier.

Best regards,

Arvind

Okay Arvind - great.

That tick after one minute (with an SFS of 60 that is) is my bad. I am working on solving that ...

Best regards,
Peter