Soundqualitiy of XX

[Flecko]

Hi Peter,
XX is as far as I know the best sounding player, thank you for that and congratulations! It has a better resolution and staging than PowerDVD, the sound is somehow tidied up. The difference is clear audible, it is not that way that I think PDVD is unhearable any more but XX is definitive better. One thing is still interesting, althogh XX playes everything from ram, it still gains soundquality when the track is loaded on ramdisk before. I use engine 2 without upsamlping (doesn't work on this PC, could it be to less ram?(512MB)).
Greetings Adrian

[PeterSt]

Thank you Adrian,

Well, my XP playback machine has 512MB, and I play it Double all the time. But I suppose you talk about Windows2000, and on that one I don't know yet.

Your DAC can do 88K2, right ?

[Flecko]

Hi Peter,
on my other PC I have w2k too and there the upsamplng works...But I think it is not so important to me because my dac sampels to 24/96 anyway. Will there be a support for 24/96 or 32/96 files? These always sounded the best on my system.

I have made some tests with the two engines. Both without upsamling. Engine 1 sounds more right. A typical thing on what I judge PC software  is how the highs sounds. Often when the player is not so good, the highs sound broken, the sound seems not to be one piece, it is diffuse. And this is what engine 1 makes better than engine 2. The result is that E1 has a "3D landscape" where E2 is mor 2D. E1 is more detailed and at the same time more plesant to hear, smoother.
Both enigines gain by plaing from ramdisk. As far as I know there shouldn't be a difference. It is not easy to hear but I am sure there is.

Greetings Adrian

[PeterSt]

Well, my XP playback machine has 512MB, and I play it Double all the time.

Oops, this is wrong. I have 1GB ...

[PeterSt]

I use engine 2 without upsamlping (doesn't work on this PC, could it be to less ram?(512MB)).

Okay, when I load a track that's too large to fit in memory (I think "just" too large) with Engine#2, it kills the sounddevice, hence no sound without further warning.

[PeterSt]

on my other PC I have w2k too and there the upsamplng works...But I think it is not so important to me because my dac sampels to 24/96 anyway. Will there be a support for 24/96 or 32/96 files? These always sounded the best on my system.

I have made some tests with the two engines. Both without upsamling. Engine 1 sounds more right. A typical thing on what I judge PC software  is how the highs sounds. Often when the player is not so good, the highs sound broken, the sound seems not to be one piece, it is diffuse. And this is what engine 1 makes better than engine 2. The result is that E1 has a "3D landscape" where E2 is mor 2D. E1 is more detailed and at the same time more plesant to hear, smoother.
Both enigines gain by plaing from ramdisk. As far as I know there shouldn't be a difference. It is not easy to hear but I am sure there is.

Adrian,

I have been thinking about this very long;

First of all I 100% trust your judgement on the "highs broken", just because I recognize it. However, not from the current XX version ...

Although it is very hard to estimate what *is* going on in your system, I think in the direction of sibilance created by your amplifier.
Please note that things get *very* complex when it is about how matters interact with eachother, react to eachother, and destroy eachother.

Sidenote : Engine#1 and Engine#2 are not perfect as such, which you could (or should) note as a fact.

Once your amp reacts improperly to the "sharpness" of the player, your speakers may resonate to it and amplify it once again. That would be an example of even over-amplified sibilance (which for sure makes the sound digital).

The most common (to my experience) cause of highs really sounding broken (like with empty holes in it) would be caused by jitter. Thus, if you are sure your amplifiers are okay (and are not a Class-D ...) I would put my bets on that.

If your DAC can resample to 96Khz, personally I don't trust that. Note though that I don't have listening experiences with that, but it would follow theories I am not in favour of. So what does your DAC do ? oversample 256 times and then back to 96 ? Or is it a non-oversampling DAC which can do it anyway for 2 times ? Mind the latter, because it can't be true. I can when it samples at 88K2, but 96 is impossible.
So again, I think your DAC oversamples (256 etc. times), and then all is destroyed. I mean, all is destroyed what XX tries to achieve ... 

In the future 24/32 bit data will be supported by Engine#3 (thus on Vista) at 88K2 or 176400 samplerate.


The RAM thing ... I don't think it is possible, unless indirectly you caused anomalies in the system. Example : when you don't play from RAM, you play from hdd, but it is (this time for me) not guaranteed that you have enough memory, and that during the process virtual memory can get active (which means access to the hdd again). When you play from RAM even then (but in a different way) you can get virtual memory alive.
It can only sound different if your system is not okay (like in fact too few memory).
Note that whether the track is copied from hdd or copied from RAM, once that has been done, it plays from the same memory. UNLESS you took away the normally chosen memory by means of creating a RAM disk (that this could sound different is theory only, but I bet it can). So if you test this, always leave your RAM disk specified, whether you use it or not.

Peter

[Flecko]

First of all I 100% trust your judgement on the "highs broken", just because I recognize it. However, not from the current XX version ...

Although it is very hard to estimate what *is* going on in your system, I think in the direction of sibilance created by your amplifier.
Please note that things get *very* complex when it is about how matters interact with eachother, react to eachother, and destroy eachother.

Sidenote : Engine#1 and Engine#2 are not perfect as such, which you could (or should) note as a fact.

Once your amp reacts improperly to the "sharpness" of the player, your speakers may resonate to it and amplify it once again. That would be an example of even over-amplified sibilance (which for sure makes the sound digital).

My amplifers have no silk highs and are not so nice to listen to like a tube amp but they are of good quality and accurate. I don't think it is depending on the system in this case. One thing  that makes me abel to hear slight differences between settings is my tweeter(ScanSpeak Ringradiator). It has an incredible resolution, a hughe bandwidth and very low distortion. I tried to speare everybody a long sounddescription so it could be a bit missunderstandable. I don't wanted to say that E2 is bad, E1 just sounds better. There is more air around the speaker, deeper stage, finer sound. Everithing sounded like there must be something better with this engine. Do you think it is the opposite and E2 sounds better (without doubling)?

If your DAC can resample to 96Khz, personally I don't trust that. Note though that I don't have listening experiences with that, but it would follow theories I am not in favour of. So what does your DAC do ? oversample 256 times and then back to 96 ? Or is it a non-oversampling DAC which can do it anyway for 2 times ?

It first samples to 24/96 and the signal gets reclocked by a masterclock and then comes the digital filter DF1704 with 8X oversamling. Beside it is  debateable if this is good or not, I just can say it sounds superb. If a nos dac sounds better I never tested ( I am a not sure if it will).

Mind the latter, because it can't be true. I can when it samples at 88K2, but 96 is impossible.

I don't understand what you mean by this...?

So again, I think your DAC oversamples (256 etc. times), and then all is destroyed. I mean, all is destroyed what XX tries to achieve ...  Happy

But a better signal still must deliver a better sound, I mean when the engine is better it still should sound better...if the dac is not too bad.

In the future 24/32 bit data will be supported by Engine#3 (thus on Vista) at 88K2 or 176400 samplerate.

Very sad that it don't will support the uneven multiplikators

The RAM thing ... I don't think it is possible, unless indirectly you caused anomalies in the system. Example : when you don't play from RAM, you play from hdd, but it is (this time for me) not guaranteed that you have enough memory, and that during the process virtual memory can get active (which means access to the hdd again). When you play from RAM even then (but in a different way) you can get virtual memory alive.
It can only sound different if your system is not okay (like in fact too few memory).
Note that whether the track is copied from hdd or copied from RAM, once that has been done, it plays from the same memory. UNLESS you took away the normally chosen memory by means of creating a RAM disk (that this could sound different is theory only, but I bet it can). So if you test this, always leave your RAM disk specified, whether you use it or not.

Good point. Ill will make the test again and consider this.

[PeterSt]

Hi Adrian,

Mind the latter, because it can't be true. I can when it samples at 88K2, but 96 is impossible.

I don't understand what you mean by this...?

"I can" should have been "It can".

This was about the not *over*sampling (like 256 times) combined with 2 times. 2 times 44K1 = 88K2 and not 96. So :
In order to get 96, first oversampling to a common denominator must take place. By that time, all the beautiful preserved quare waves have become pure sines.
2 times upsampling is just allright (to unquaree thee highest frequencies).

When quared (or squarish) waves come out as sines, that's nothing else than distortion.
That's why.

Lastly, Engine#1 and #2 just *are* different. So both do sound differently, I know that. From this follows at least that both can't be 100% right. And I know both for sure are not right (a chance #3 still has ).
I say that Engine#1 is better, BUT, it can't do the Doubling. Since both are not perfect, #2 with doubling should workout (!) better. Theories ...

Btw Adrian, my remarks about your remarks on the highs are not in defence of the player (I mean that), but are in the area of trying to improve your system where possible. Maybe for fun you could try a nos-DAC. Then try to find some synthesizer music you like (can you ? ). That will show the most explicit what's going on.

Regards,
Peter

[Flecko]

This was about the not *over*sampling (like 256 times) combined with 2 times. 2 times 44K1 = 88K2 and not 96. So :
In order to get 96, first oversampling to a common denominator must take place. By that time, all the beautiful preserved quare waves have become pure sines.
2 times upsampling is just allright (to unquaree thee highest frequencies).

The upsampling is bevore the oversampling. I red an articel about it and the autor said that the asynchron upsampling (from 44.1 to 96) is better than upsampling from 44.1 to 88.2 because of jitter reduction. He didn't proved it mathmatical or somthing like that but I guess thats what is written in the standard books of DSP. Perhaps there are other aspects, the articel was not very detailed. To multiply with an even number seems to make sense, but it could be much more complex.

Btw Adrian, my remarks about your remarks on the highs are not in defence of the player (I mean that), but are in the area of trying to improve your system where possible. Maybe for fun you could try a nos-DAC. Then try to find some synthesizer music you like (can you ? dntknw). That will show the most explicit what's going on.

Comparing to a nos amp would surely be interesting, but I would not build or buy one for that, perhaps sometime I know somebody next to me with a nos DAC I like elektronic music, do you think it is good for testing because of the simplicity of the synthi-sounds?


Regards,
Peter

[PeterSt]

I like elektronic music, do you think it is good for testing because of the simplicity of the synthi-sounds?

No ... because of all the square and sawtooth and triangle waves in there.