[svyr]

>Try minimizing any buffers in your system and then stress it with some CPU hungry application and you will probably hear some drop outs, which are an extreme example of timing error.

buffer under-run is not jitter. and it's not a timing error. It's a buffer underrun - your CPU or IO are too busy to fill the buffer (either read the file from disk or decode the file)

 

[MaciekN]

You are right, I just oversimplified and mixed few different things. Anyways, it's explained in the article.

 

[Dradder]

MaciekN -- I'm glad you found my posting to be of value. If the postings from certain other people aren't of value, it's because they actually have nothing to say regarding this subject. That they continue posting in this thread saddens more than it provokes, but it's their choice. Best wishes.

 

[svyr]

in case you guys feel like restarting http://www.notquitewrong.com/rosscottinc/2011/08/03/so-youre-mad-about-something-on-the-internet/

 

[Dradder]

Didn't even feel like starting, let alone restarting. Now that the people who actually didn't have anything to contribute to the discussion have dropped out, everything seems to be going along groovylicious! Cool, huh?

 

[Dradder]

MaciekN -- Further on the admittedly subtle and tentative results I'm getting with my Game Booster+ protocol. The effect, as far as it might or might not exist, and might or might not be audible, seems to be most noticeable in test tracks with a lot of rapid initial transients. For example, another track I use for evaluation purposes is a song titled "Ugly" by the Korean pop group 2NE1. The drum kit on the track includes what sounds like a parade snare; with the GB+ protocol, the rattle of the metal wire snare on the bottom of the drum seems a little more audible. (Again, this is a very subtle difference, not a night-and-day thing.) I'm not sure if this is because the protocol is improving what's known as the rise time of the snare's initial transient sounds, or whether the protocol is reducing some audible element that was in some way obscuring or masking the initial transient sounds. Very subtle, and I'm still evaluating whether it's genuine or not. But it's interesting and enjoyable, which is all that matters.

 

[MaciekN]

Okay, so the CD transport has arrived: it's Harman Kardon hd750. Testing in progress, I'll try to post on it later.

 

[MaciekN]

I think I have pin-pointed the differences with Game Booster, they are definetly audible. I was switching between PC and Harman, writing down the differences with or without the Booster vs CD. I'll give detailed results in a longer review on my findings, maybe tommorow.

 

[Dradder]

Quote:

I think I have pin-pointed the differences with Game Booster, they are definetly audible. I was switching between PC and Harman, writing down the differences with or without the Booster vs CD. I'll give detailed results in a longer review on my findings, maybe tommorow.

Super. Looking forward to it.

 

[MaciekN]

The contenders were:
PC with foobar without any DSPs (except resampler) set to Kernel Streaming, s/pdif conversion done by integrated soundcard ALC883, connected to external DAC via optical cable.
Harman/kardon hd750, a CD of which I know little except that it has digi out, connected via coax
The DAC was Beresford 7520 with DAC chip replaced to wm8716, input caps replaced with nichikon KZ, opa2228p in headphone amp. Headphones used were recabled and reshelled Alessandro MS-1i
CDs used were: Mike Oldfield's Tubular Bells III, Madonna's Ray of Light, and Ultrasone Reference CD.
 
In the order of playing
 
Tubular Bells III
The first test song from this album was Jewel In The Crown, consisting mainly out of electronic drums and some effects.
At this album Game Booster was disabled, so no "sound enchancers" applied yet.
Harman gave a very clean, slightly shy but deep bass, which sometimes sounded fuller than PC in spite of it's thinnes.
With PC there was more bass quantity and it was much more of a mid-bass rather than the lower one. It did not sound as clean as on Harman
 
Then I played Man In The Rain, vocals along with some guitars.
With the CD all the echoes were easily captured, with singer situated deep into the soundstage, so obviously there details were much easier to capture here. Overall soundstage had more air to it.
In comparison, PCs vocal sounded warmer, with some more body. There wasn't even half of the reverbs heard from the CD player, singer sounded as if she was pushed to the very back of soundstage and put in a small room without echoes.
Overall, on this album, PC sounded warmer with clearly less detail, duller.
 
Next was Ray of Light from Madonna, so vocals with some electronic effects and deep bass.
Cd player's bass was again clear, this time even and rhythmic, as well as deeper than from PC.
Next to Harman's bass, the one from PC was more boomy, there was something unpleasant to it, something scratchy.
And here I used Game Booster. Previously, apart from foobar the only visible applications opened were PC Probe and antivir but Booster stopped 24 other apps nontheless and defragmented 50 mB of RAM (yeah).
Now things changed, the bass from PC was no longer boomy and scratchy, it had similar to CD's quality to it but there was slighly more of it in terms of quantity, so it was just as deep and a little louder.
Next were the vocals:
Harman again placed the singer deep but this time it was impressively deep. Her voice was a lil quiet but very clear. Overall I rated it more musical and more speaker-like in terms of soundstage.
The vocal from PC was pretty closer, not as clean but only a little. It was more dead-on-spot, there was a strong feeling that she was right there, in that one point from where she's singing. In comparison, the vocal from the CD player seemed as if it's floating in the deep outskirts of soundstage.
At this point I noticed that PC had more around-your-head presentation, while the CD's was very nice, deep, yet narrower but still with an overall feeling of being larger.
 
Now, the Ultrasone Ref CD, it has a very diverse material, some classic, orchestras, individual live instruments, and effects like the sounds of sea, birds, fireworks, everything binaurally recorded.
At fist PC was not supported by Booster. Bass on organs was weaker, not as deep as the one from CD, with more of a left to right soundstage, nowhere near as deep as Harman. Details were much harder to pick, if I heard something new it was with the CD as source, only then I noticed them from PC.
Then I turned Booster on. The bas went as deep as CD but there was less of it in terms of quantity, soundstage become even more left to right with clearer localization of instruments. Details were now on par with Harman but highs sounded slightly but noticably muted.
Overall Harman had a sense of freedom to it's sound, it gave an illusion of larger space and a joyous, unbound quality to the instruments, with clearly better highs.
PC had a more forward presentation and firmer localization of instruments, the sound could be described as aggressive and more meaty but overall I prefered the freedom of Harman.
 
 
So to sum up, without Game Booster my computer transport was weaker than a nothing-special CD, rated good but nothing outstanding in it's class. With the Game Booster it became a matter of preference, as neither was clearly better than the other. Thanks to Game Booster frequency response was better at extremes, sounstage became firmer with very nice localization, and details were easier to pick. I'll try to test Fidelizer in the same manner soon.
 
WARNING: Game Booster may in some cases mess up your system. I have encountered no problems so far but use it at  your own risk
 
If properly configured computer transport can easily compete with 1k$ CDs or even higher, then I have a lot to improve in my case. Gapless playback with no drop-outs is only the first step, as now speaks my experience. Disabling unneccary services does improve the soundquality
 
 
 

 

[Dreaming Of A Better ...]

"PC with foobar without any DSPs (except resampler) set to Kernel Streaming"
 
you're not getting a bit-perfect signal if you use DSPs in foobar.

 

[MaciekN]

Yep, but I wouldn't get bit perfect  beacuase ALC883 is clocked with 24MHz oscilator, so if I play a 44.1kHz file it  gets upsampled to 48 by the card anyway, so I'm resampling to 96kHz from the very begining :/

 

[Dradder]

MaciekN -- Thanks for the interesting report. Very similar to my own perceptions so far about Game Booster effects.

 

[cYbernation]

nevermind.....

 

[random007]

Quote:

 
 
With wasapi or kernel streaming or asio you surely get bit-perfect signal on the software side but there is also the aspect of timing. Even if bits are correct, if they reach the DAC lil too early or lil too late they will have audible impact on SQ, based on the size of these time shifts.
 

 
No it wouldn't. It doesn't matter about timing, the card's sound processor chip buffers data and instructions until it gets everything it needs, then it sends the appropriate data to the DAC chip (often the same physical chip). What the DAC gets is ALWAYS the same as the program (via various software layers) sends. Digital electronics has to work like that or nothing at all in your computer would work properly.
 
I'll make these points:
 
. Data is either entirely right or entirely wrong with no in-between and if an audio card was receiving bits the slightest amount differently to the way they were in memory, there would be considerable loud clicks and other noise all the time. It's the same reason there's no such thing as good or bad sounding S/PDIF link and why running an expensive blue marker pen around the outside of a CD was pointless: digital signals work or they don't.
 
. Only a few data paths within a PC are error-corrected, the rest rely on data getting to where it's supposed to in the exact same state as it left. Computers are designed to be very good at that because they'd stop working if they didn't. The requirement is for perfection.
 
. The DAC does not recieve data directly from the PCI/PCIe bus, the controller chip organises what the DAC gets and when it gets it to ensure perfect timing.
 
. All the data bandwidth involved in making a stereo audio signal is millions of times smaller than other vital functions. In the world of digital signals, an audio signal is not a very challenging task, quite the reverse. 
 
. You need to be saturating the processor or (more likely) storage medium (eg hard drive) to cause a drop-out. Hardware buffer under-run manifests as series of clean drop outs or a rapidly repeating sound, not a gradual reduction in sound quality as CPU<->memory<->bus<->s/card transactions slow down, which should be the case if audio quality was affected by system use.
 
. Unlike analogue signals, digital signals don't know about sound frequency or THD and if a DAC thinks there's a 1 when there should be a 0 (or visa versa) there's an equal chance of a wrong Most Significant Bit as any other, which would make an horrendous noise. Any random bit jitter will cause horrendous noise, not a consistent change in tonal quality.
 
I'm comfortable in saying system activity will NEVER (apart from drop-outs in extreme situations) affect audio... because it's physically impossible. If it were the case, computers wouldn't work. Claims that a system sounds different depending on 'background tasks' should be taken with a pinch of 'placebo' flavoured salt (never underestimate the power of placebo!), especially when words refering to tonal character and frequency bands are used. Again, it's just not possible.
 
Oh, and that transports article ascribing tonal characteristics to digital jitter in a D to A process: highly problematic (ie, bs).
 
[edit for readability]