0.9z-4-0

[Flecko]

Just to explain what I think. I don't want to say throw z4 away. It is a new approach that is promissing. But to get someday a xx1.0 somewhere should be a cut. I think z2 is good point to do that. There could be some sort of debuged Final Version of xxz2, that marks what has been achieved. And the idea of z4 can be followed to get to xx2.0. Z2 Final Version could be used as some sort of reference, where to go and then go beyond. It will be also much easier for new people to join xx. This would be a version, where most people agree of the set up (adaptive, scheme3). So if a new guy comes, he just has to copy paste the settings and he will probably get a very good sound. This would be some sort of island where we can savely return after exploring the new stuff.

[boleary]

When I switched back to Z2 after a month of Z3 I was really glad for the less forward and less, somewhat harsh sound, though z3 still sounds really good. But z2 has an overall less articulate sound in all ranges that changes dramatically with changing the SFS. One can make Z2 "sparkle" with lower SFS but not, on my system, in a smooth, pleasing way. Large SFS in z2 just make all frequencies more softer and more muted; I would always look for the balance between the two for the best sound, which changes a lot with different albums. I'm afraid I probably erred on the more muted side, keeping SFS between 95-115 but that also allowed playing music a lot louder. With Z2 my focus was to eliminate harshness, which I was able to do, but I'm pretty sure the sound was still not "right".

Z4 with my current settings is head and shoulders above z2 for clarity, richness and smoothness; I'm amazed at just how much more layered the sound has become. And, as someone posted before, I can play with around 10% less volume.

Sometimes I'm not sure how Peter moves forward with such very different experiences of XX!

[PeterSt]

And, as someone posted before, I can play with around 10% less volume.

I read this too, but I read "through" it. Didn't know what to do with it much. But now I read this again in this post -and at a first read I thought you just repated it- and now it keeps being in my head.
Here it is the same ...

As we may remember, before we were always saying "and hey, I can play much louder with this version" ... and it was a virtue.
Now we "can" play less loud, and it is a virtue ?

But what actally is this ?

I know from before - and it is a kind of logic - that being able to play louder was about less harshness. At least here that always was a key to that. Today - and I explicitly watch that - this is no issue anymore. Instead, now it seems that there's a limit which is a more natural one : "this piano doesn't play that loud in real life" or something ...

But I am fairly sure there is also something with the bass;
I can't tell how that is with others, but if I will be receiving more bass in a next version, I will really start to wonder how it is possible;
I was at a concert a few days back, and it was there that I noticed that the poor guys were not even able to have a more loud and stiff bass than I can do in my listening room. I tried to watch the level, and could not imagine that it was really more loud than in my room. The place was quite bigger allright, but still. Before I have always been so jaleous on those basses guys, and which is so easy for them having their dedicated speaker(s). But not so anymore as it seems. I am now outperforming that ...

Somehow there is a bottom end (but it starts fairly mid) that has much more output than before. For a great deal it will be W7-SP1 related, but the "remainder" seems to come from Special Mode which jus has everything so much more under control. That same Special Mode which before emphasized details so much that the foot tapping went away. It emphasized the highs with it. But not so anymore. Something really REALLY starts to work, and I guess it can't be explained otherwise than magic to the sense of pieces of the puzzle starting to fit. Coherence ... ?

Long story short, and not sure where it comes from exactly, here the volume "can" be lower too. It is not because otherwise the music annoys; it just happens ...

Peter

[Flecko]

Z4 with my current settings is head and shoulders above z2 for clarity, richness and smoothness

For richness and smoothness I would say yes. Clarity... depends on what you mean. What I like on z2 in the highs, is it's "clarity". Voices have more presents. It is more foreward but in a good way. Also I think there is more space between the instruments. Z4 sounds smoothed and a little unnatural compared to z2. For example, with z4 the overtones of a steel guitar are more diffuse. Z2 gives a cleaner focus.

[PeterSt]

Hey Adrian !

I try my best to follow what is happening at your end, but it is a bit difficult for me;
I think I already implied some apples and oranges at your side, so asked "how and what". Even if you gave the answer, I guess you need to repeat a little more what you exactly compare. RAMDisk seems crucial here. The fact that you "need to" upconvert to 96 from 44.1 ... I exactly never heard that would do some good.

Apart from it all, you mostly are "out of line" so to speak (only meaning : usually you have a different oponion from all of the others). With this I want to emphasize that it is *not* like "thus how can you be trusted ?". The contrary !! but now I really need to judge how you judge or otherwise it can't tell me much.

I hope I can make myself clear a bit !
Peter

[Scroobius]

These are my experiences with z4 but I am sure that using a laptop with probably not enough resource has a bearing.

Firstly z4 is a big improvement over 3 to these ears and enough to keep me up late for a couple of nights going through my collection. More detail would be the main difference. Vocals particularly sound much improved with much more articulation but everything is better. Sometimes the sound can be a tad hard but there is no going back to 3.

Settings are:- Adaptive & 1024. Special never used to work (resource issue probably) but it does now however there are regular "clicks" which improve (but don't disappear) when the buffer size is increased. Mixed memory arrangement seems to work best but it is difficult to say if the differences between the 3 memory settings are that significant. Certainly compared with the improvement from 3 to 4 the differences due to the memory settings are small.

Great improvement.

[crisnee]

Flecko Said "I tryed to started a project like this here. There was positive response but nobody except me uploaded something,  You can have my testset or share some files of your library"

I remember reading that and commented at the time. What I'm suggesting is similar but possibly simpler (see my post earlier in this thread). The question is can one attach a .flac file to a post and if so does Peter allow it (is it possible do you allow it Peter?).

If it's possible, people could just add a file at any time (no name, tag info erased, and only one track from a particular album to avoid pirating/copyright issues) and maybe explain what they listen to/for in particular and what they hear. It could be an ongoing thread and people could contribute at anytime.

Chris

[PeterSt]

Thank you very much Paul. One thing for your own improvement at this moment :

but it does now however there are regular "clicks" which improve (but don't disappear) when the buffer size is increased.

Please don't use this (Special Mode), if this is happening. You actually won't know what you'll miss more than some skipping parts which "tick";
Think about the number of samples you will be implying (latency), and how often something may be skipping per second without you noticing it.
I kept away from Special Mode for as long as ticks happened, and now they have gone here, it suddenly sounds marvelous at the same time for everything and all. Can be coincidence, but I don't think it is.

So, be careful !
Peter

[Flecko]

I try my best to follow what is happening at your end, but it is a bit difficult for me;
I think I already implied some apples and oranges at your side, so asked "how and what". Even if you gave the answer, I guess you need to repeat a little more what you exactly compare. RAMDisk seems crucial here. The fact that you "need to" upconvert to 96 from 44.1 ... I exactly never heard that would do some good.

In my case it can be explained easily. There is only one clock in the DI and it is a 24,xxx Mhz clock. Fits best to 48,96.. Khz. Also in theory, thats what I "heard" some times, performing asynchronous upsampling reduces jitter. Using that upsampler increases the details and the freshness in the highs by an easy noticable amount. It is little aples and oranges and just my experience, but doubling with the hiface never brought an improvement for me. I know doubling works for a lot of people. I think this could also be explained because of my hardware. The Ref7 dac uses a PLL and has also a 24,xxx MHz clock. I don't know if this clock is used with the pll or what their purpose is. Just an Idea why it sounds like that. But concerning what I hear, asynchronous upsampling can not be that bad.

Apart from it all, you mostly are "out of line" so to speak (only meaning : usually you have a different oponion from all of the others). With this I want to emphasize that it is *not* like "thus how can you be trusted ?". The contrary !! but now I really need to judge how you judge or otherwise it can't tell me much.

Thanks!
Ok I explain:
I don't use Ramdisk for comparison.
I have both versions, z4 and z2 on my D: partition of my only Harddrive (Samsung F1)
I start both versions at the same time.
They have exactly the same settings (look at my signature), except that z4 uses clock=12ms and MC mode
I load in every xx only one and of course the same track.
Than I play the track with one version. After some time I stop and play from that exact position with the other version (For example at minute 2 second 45).
Sometimes I just hear the first 20 seconds of a track and start the track again with the other version.
Then I take the version, that I think is better and hear for a longer time.
Later I compare again to the other version.
that all can happen quit often.

I feel not offended at all but just to remind you. Last time I was out of the line, I told you that vista is not "by far better than w7". Which turned out to be true. You can argue, that I said "no difference" or "more or less the same sound". But you can imagine what expectations one can have, if everybody sais "by far better". And I already had some of the SP1 updates due to auto update.

It is always difficult to judge something objective, that is essential a subjective thing. But I do my best
Z4 has that niceness (which could be just smoothing or jitter). I can totaly understand, why somebody would like it. I like it too (but not as much as z2). I just want to indicate, that there could be something wrong with it. Z2 sounds more right to me. More presence, cleaner. A bass doesn't sounds brown, it sounds black (most kinds of bass do). z4 sounds more golden in the highs and more brown in the middle, reminds me on my lovely but worse cd transport. z2 is more silver in the highs and black in the bass, more natural.

I just listened to "sade-smooth operator". I am pretty sure that if you could meassure the jitter of the two versions, z2 would have lower jitter. I can't help myself

[xp9433]

Flecko
Clarification please. Are you upsampling to 96kHz in XXHE and feeding this signal to your DAC, or are you feeding 44.1kHz to DAC and letting the DAC upsample to 96kHz?
Thanks
Frank

[PeterSt]

Last time I was out of the line, I told you that vista is not "by far better than w7". Which turned out to be true.

But this is exactly why I emphasized (or tried to) this not being a negative at all. I hope to show that I listen to everybody, and I am quite sure (if not just 100%) that it's not only me wondering how you get to your findings. I am confident we will all always think "couldn't that be true ?".

but then

and has also a 24,xxx MHz clock

I won't say this explains a lot, but at least it puts things in an important perspective.
I'll leave it to that for now (must think about it), but say "thanks a lot" for your ever efforts of writing down your observations. Great.

Peter

[Suteetat]

I am curious about asynchronous upsampling. Do you mean, for example, upsampling 44.1 to 96 rather than 88.2 ?
Even more confusing part is about using 24 mHz clock or do you mean using 24 multiples clock sampling ie 96 and 192
clock input or you use 24 mHz clock input and upsample everything to either 96 or 192?

Long ago before discovering XXHighend, I used mediamonkey/Lynx AES16/ Esoteric D05 DAC. I used to have Esoteric clock output
to Lynx AES16. Mediamonkey will then upsample 44.1 file to match whatever clock input I set from Esoteric. I definitely found
44.1 with 44.1,88.2, 176 clock input sounds much better than 48,96,192 at that time.

Now aday, I don't use any oversampling at all. Mainly because my current digital front end, Playback MPS-5 convert everything to DSD2
internally. So I find that there is a bit more purity to sound when I just run everything straight through to PD without additional
oversampling from the computer end. (However, that was not the case with my old DAC, Berkeley Alpha)

[PeterSt]

Do you mean, for example, upsampling 44.1 to 96 rather than 88.2 ?

That is what Adrian will mean yes, although indirectly, because "asynchronous" is not related to the "not even" sample rate as such.
But, what Adrian will do is upsample to a rate which is best for his clock, which is an 24.xxx only (and which is the base for the 48KHz sample rate base really). So, theoretically indeed it will be better to convert to the base of the clock (if there is one only, like in Adrian's case), but as I said, I have never heard of any means of upsampling being good (for SQ) in the first place. Well, not in here, and I am not talking about "Foobar people" using whatever Rabbit upsampling and who can only hold that against Foobar itself (and I skip Arc Prediction, which is what 99,9% of people like ... ehh, not Adrian).

In my opinion (though a rather dangerous "conclusion") it will be the reason why Adrian is so often obtaining different results from others. I mean, he *has* to use some means of upsampling because otherwise his DAC won't perform in the first place (it would be NOS/Filterless), and even if he doesn't himself, the filtering options in his DAC will do it for him. And both are not good ...

... which doesn't definitely say at all that "not good" is an absolute measure. So, if his DAC sounds good to him (meaning better than anything else for that price), why would there be a problem. There is one though : we can't compare results.

I am the most serious : Of what I can tell Adrian is a very good observer, but many more in here are. There has to be a reason why he judges differently, and I refuse to believe it is because of e.g. me bing less capable. But I guess the answer to this has just been found.
But keep in mind : dangerous for a conclusion without even being able to listen (but I also guess we all get rather experienced in doing so, which is mainly based on trust, and foremost : honesty).

Peter

[Flecko]

Even more confusing part is about using 24 mHz clock or do you mean using 24 multiples clock sampling ie 96 and 192

If it is as I think, then it is because 24.xxxMHz devided by 48khz or 96kHz is a even number. Devided by 44.1 or 88.2khz is not.
The signal leaves XX with 44.1/16 and is ubsampled im my usb/spdif converter.

here is an article by benchmark about asynchronous upsampling. I didn't had time to read it myself. Maybe there is something ineresting for you/us. (I will read it later)
http://www.benchmarkmedia.com/discuss/feedback/newsletter/2010/07/1/asynchronous-upsampling-110-khz

There is one though : we can't compare results.

I think it depends on what we compare. If it is using arcpred or not, doubling or not I would say you are right. So everything related to this technical point can not be judged in an absolute way. But when it comes to computer related things, like different versions of xx, it should work. If it sounds better, it sounds better. As long as the hifi system is good enough to let you hear whats right or wrong. Otherwise, you could not compare with anybody not having the same setup. Which is a problem we have for sure, but somehow it works...

But I guess the answer to this has just been found.

It might explain something but I don't want to give up the z2 thing yet. Maybe it would help, if we first understand why Ramdisk doesn't sound good. Or why the smoothing ramdisk does, can be preceived as good but is not. Here we have theories, which I don't know (maybe you can explain it peter?). I think a similar thing is going on with z4. Then z4 and z2 could be compared. But it should be done and if you say it is not good, it is fine for me.

[PeterSt]

Maybe it would help, if we first understand why Ramdisk doesn't sound good.

Bingo !

Maybe it was a typo, but at this moment I rather say :

Maybe it would help, if we first understand why without Ramdisk doesn't sound good.

Or why the smoothing ramdisk does, can be preceived as good but is not.

Looking at this outside of all further context ... the "is not" is my own idea only. I only want to say : be careful not to copy my ideas as long as theories are not really theories, but empirical findings only - next trying to find theories for that.
It can just as well be that (e.g.) I have to replace my speakers !

If it is as I think, then it is because 24.xxxMHz devided by 48khz or 96kHz is a even number. Devided by 44.1 or 88.2khz is not.

I don't think this will be related to even numbers (upsampling "uneven" is almost as easy - and sure is as harmless as "even" upsampling (for normal filtering means that is; not for something like Arc Prediction which can do "even" only)).
The point is though : a 44.1 sample rate should not be dealt with by a 48KHz based clock. Physically it can though, but samples will be missing here and there (so that won't be the case I'm almost sure).
In the end (how I see it) it is about one thing only : it won't go without upsampling, and *now* you are stuck to it. To what degree this harms by itself completely depends on the algorithms used in the SRC (DSP in your case), but one real downside is clearly visible already : you just can't use Arc Prediction. This alone is enough to perceive the sound totally different from others, because :

a. Almost everybody uses Arc Prediction (out of free will );
b. a determines that "something" should be done to redbook, or otherwise nobody would use it;
c. this thus counts for you too;
d. but you can't.

See ? your destiny has been determined. haha

But for some counterweight :

Ad b.
It is not said at all that Arc Prediction is the only good means to indeed do something about redbook. Otoh again : think;
Arc Prediction will overrule all the numerous means in all DACs all using this wrong means of upsampling (I leave NOS out for now). So, just because of a. above is happening , it kind of prooves that all those oversampling / SRC etc. means do not work well (by now this is just statistics). So, why would yours suddenly be a good one ? it can, but chances are not large. And IMHHetc.O they become zero when a DAC (or whatever it is in the chain) has an oscillator for one sample rate base only.
But maybe I'm thinking strange ...