The same way magic power cables affect sound quality.  
 
If upsampling is not being used and an application is "bit-perfect," there should be no issue unless for some reason there is jitter specifically associated with the app.  However, using a device like a HiFace, or reclocking circuitry in the DAC will eliminate this variable anyhow.  
 
Of course, there's always someone who believes they perceive a difference, and short of A/B testing (which will never happen), there is no way to say for sure that someone in fact can't hear a difference.  With due respect to Bixby, I would question his ability to hear a difference under truly blind conditions, but I have to freely admit that he could be 100% right and there's something not obvious to consider or that I can't hear.  Computers ARE very complex, though beyond jitter and upsampling I am not aware of any material consideration that affects SQ, it doesn't mean it's impossible.  My $.02 is it's just not likely.  
 
So my real point is that everyone should listen to the players, as most are free or have demos, and if possible really try to separate what they EXPECT to hear from what they CAN hear.  And that's really hard, as our expectations absolutely do affect our perceptions...
 
That said, I'm generally pretty sensitive to distortion and I honestly can't hear any difference at all between players using ALAC for redbook media, unless upsampling is involved, and then the difference is there, but it's subtle (like I want to listen longer and just feel more immersed in the music).   I have used almost all the players discussed, except Amarra.

I havent heard a difference in power cables yet, but I have no problem in hearing differences in the different players. I dont have an explanation of why this is, but there is a lot going on in a computer. A usb.spdif converter just supplies a clock. So you eliminate clock jitter, but you havent done anything with how the music is processed, or how the core audio is bypassed, or how stressed the processor is causing jitter there. You just know that all the bits have arrived and were timed by the clock in the converter. We can go between different computers  for front end and have bit perfectness and sound different and I dont think anybody questions that. Why cant software have an equal effect? Obviously we are missing some other variablres. Just curious, how do you decipher bit perfect. Does it check every bit that arrives compared to with what is on the disk or does it check against what was processed? I know some dacs have bit perfect checks, but what exactly are they checking?

I'm going to back Bixby here. You just need to have a resolving and transparent enough system to hear the differences in players.
 
I had a Leben amp paired with a Chord DAC, both very transparent and resolving components. I could EASILY hear the differences between Ayrewave & Audirvana. Before Audirvana added its integer mode, it was very mechanical and digital sounding in my setup.
 
Once I switched to an iDecco amp, not so much. It just didn't have the transparency and resolving power for critical listening to discern between components.

Either that or it had a reclocking system that was superior and overcame the timing issues. You really don't know...

At the end of the day, players are different, but mostly when they process the audio in some way, when not using Core Audio for eq or other functions... Systems with lesser reclocking would in theory be more prone to show "difference" with bit perfect audio with varying degrees of jitter.

Again, reclocking would be the answer. And good reclocking can be less than 10 pls, which is way lower than you woucl perceive...

Just theorizing on this, but I am an EE and in audio and computers, and to me this appear the only immediate answer to why people really could hear a difference on "bit perfect" output.

When I first used Amarra, it was necessary to set up Mac OS X to use the input clock as the clock source. That might have something to do with it.  Directly interfacing with the hardware and bypassing any processing by Mac OS X must be helping in some way too. It would be good if we could get more info from a knowledgable software engineer on this or if someone measured the digital signal output with suitable gear while each player was being used.

Currawong is most likely on to why we hear differences.  It is most likely how the program reacts with our hardware, which stacks are involved and which are bypassed.  It is also interesting to note that differences in sound occur when programs are running along with the player and when they are not.  System resource use while playing a file may be another variable.  
 
I do not think differences are necessarily caused by clocks although async devices with their own clocks actually make it easier to hear differences in players for me, probably because of the increase in timing accuracy and lower jitter.  I think a lot has to do with system noise and perhaps some type of quantization noise.  At least when you use a volume control on any of these players you are introducing a BIG variable.  That variable is how the volume control works in each player.  In some, you throw away resolution and you do hear that difference, in others you add dither  and their are many ways to dither with different algorithms that actually do sound different.  They are quite audible on a highly resolving system.  
 
I am not going to work too hard on understanding why players sound different, I just accept the fact that they do.  That being said, I also have heard audible effects in changing some power cables and lots of folks do not.  I don't have a problem with that.

This explanation makes no sense to me digitally, in bit perfect mode, which was what I thought we were discussing?  I have been saying that differences will exist as soon as you modify the signal, and I guess we're on the same page.  I think your comment on re-clocking is probably exactly correct when considering non-bit-perfect playback (e.g. upsampling).  Eliminating the jitter should reveal the qualities of the scaling and shaping algorithms more clearly...
 
That said, once a signal is digitized, there's no such thing as added  "quantization noise" if it's bit perfect.  So dithering with noise to cover quantization from digital volume adjustment or EQ/effects is definitely going to vary based on if its Core Audio algorithms, or proprietary.  If Core Audio, they should be identical between apps, but if proprietary will certainly vary.  Again, if bit perfect, there is no quantization or dither being added, so you are simply reduced to jitter as the variable.
 
"System noise" would be analog that affects the power supply of the DAC, but it can't modify the 1's and 0's, except possibly as jitter.  So any software purporting to be bit perfect will put out the same bits, and reclocking will eliminate any possible electrical difference unless there are uncorrectable transmission errors, which usually result in signal drops or audible problems.  
 
Don't take this wrong, I'm kind of clowning around because the word-play is cute:  When it comes to suppositions about audible differences, you are OK to believe so you hear, while I don't believe so I don't hear.  Neither approach is absolutely right, and both approaches have limits, so as usual the truth will be in the middle, and in reality we both probably follow both approaches to a point.
 
That said, I think the main advantage to my approach is that it saves me a lot of money on power cords and cables unless I can see (and usually measure) the benefit, and as many claims are engineering nonsense or utterly unprovable, I seldom am tempted.  Being in the audio business made me a cynic about that stuff...  
 

 
 

If you have a source which re-clocks everything then why would it matter the clock functionality of the program you use.  Not every DAC does this, but if you do have one, how would the sound change from program to program?

I agree with Bixby that by eliminating jitter, you're going to reduce one of the more obvious variables in the digital chain.  For any player that's not running bit perfect (using digital volume, EQ or upsampling, for example), you will be better able to hear the effect of the processing because the overall system will be more resolving of fine details.

Quote:

mrspeakers said:

I agree with Bixby that by eliminating jitter, you're going to reduce one of the more obvious variables in the digital chain.  For any player that's not running bit perfect (using digital volume, EQ or upsampling, for example), you will be better able to hear the effect of the processing because the overall system will be more resolving of fine details.

Click to expand...

I'm sort of repeating myself, but what about DACs which reclock everything which comes into it?  Would the application used for the files be relevant at this point in terms of jitter correction.  

Quote:

mrspeakers said:

I agree with Bixby that by eliminating jitter, you're going to reduce one of the more obvious variables in the digital chain.  For any player that's not running bit perfect (using digital volume, EQ or upsampling, for example), you will be better able to hear the effect of the processing because the overall system will be more resolving of fine details.

Click to expand...

 
Quote:

davidmahler said:

I'm sort of repeating myself, but what about DACs which reclock everything which comes into it?  Would the application used for the files be relevant at this point in terms of jitter correction.  

Click to expand...

 
Is it even possible to eliminate jitter entirely? Or is it more realistic to try and reduce it as much as possible, one stage at a time? I don't know, but I doubt any re-clocking system is "perfect" as far as jitter elimination is concerned, and I'm pretty sure "bit perfect" is another thing entirely. More like you're getting your song file processed into analog in the same digital form that the file was in when it is played at the transport level. Not upsampled during conversion, for example.
 
If the bit rate, sample rate and bits per sample are all unaffected, it must be "bit perfect", no?
 
I think you can EQ digital music any which way you want, it doesn't affect the bit rate so the music will still be bit perfect, just equalized for playback purposes. Just like in the analog realm, there are good digital equalizers and bad ones; I personally avoid them all. Adjusting the volume is just attenuation; It can affect the FR with a limiting effect if you do it in the wrong place, like on your digital source.
 
But I could be wrong

A good re-clock will essentially have such low jitter it's not possible for their to be jitter artifacts.  CEntrance claims 1 picosecond of jitter, which is 10 to the -12, or a trillionth of a second.  Since the audio sample rates are measured in tens to hundreds of thousands of samples per second, this is to all intents and purposes, eliminating it.  There are many other products that have low jitter, others ignore the issue completely or rely on a cheap chip to regenerate the clock, so the mileage will vary.  
 
Since jitter is pretty universally accepted as "bad," looking for DACs with low jitter is always a good consideration factor, though not the only one...
 
Lastly, "bit perfect" only means one thing: the 1's and 0's in the playback stream are exactly what was in the audio file.  There is no change due to any adjustment by volume, oversampling, etc.  This is not the same as "time perfect," where the data doesn't arrive regularly, which is jitter.
 

True "bit perfect", to me, means that the bits are picked up from the media (CD/HDD) and moved to the DAC with no change.  Since the 0s and 1s are transmitted via a wave analog signal across whatever interface, I do think Jitter can matter, as the frequency that the 0s and 1s get moved along and read by the DAC can be non-constant, or different than its supposed to be.  It is up to the DAC to convert the analog signal from what is sent to it, and convert it to true 1s and 0s.  Again, timing can be an issue.  Some of the more recent DACs do this much better than was available before, and using a single clock to control the frequencies at both sending and recieving ends is not as important as it once was.
 
Nowadays, the information stream from the media to the DAC is much more complicated, in that the OS of whatever computer you are using will do something to the sound, rescale, mix with system sound, or just paw through the data a lot on the way to the sending hardware.  Cutting out this manipulation has been the goal since I started Computer Audio several years ago.
 
Even more recent are then new software programs which actively involve data manipulations, first by keeping it out of the computer audio stream, and secondly actively resampling, and resculpting the data as it is passed along to the DAC.  Some of the interest here comes from the much more available hi resolution DACs that are popular now.  Back when your DAC only did 16/44.1 or 16/48, there was not much point in upsampling 16/44.1 files.  EQ and manipulating the data further was eschewed by the audiophile community, while popular and common in the "pro" community.
 
While there have been software players other than iTunes for quite awhile now, they mainly offered simplicity over iTunes ever growing capabilities (ie other than music playing).  Amarra broke the ice with a program that sidestepped Core Audio on the Mac, and did something to the data, other than just pass it off to the DAC.  In doing so the sound is "better" and made music listening more enjoyable.  If the price had been lower back then, and the software security less draconian, I think we would have seen the renaissance of software players that make the sound "better'.  PureMusic put the first pressure on, and Fidelia is really stirring it up currently.  I think that these three all sound a bit different, and can be configured to do different things within the programs, allowing customization for listening tastes, and different systems.
 
Between "hog mode" of both sidestepping Core Audio (while retaining access to some of the functions) and preventing other use of that DAC by the computer, combined with the  (optional)resampling by some very nice software resamplers, the new software sets a new basis for what an audiophile computer playback system needs to provide.  Prices now allow testing more than one of the players, and try them on all of you music systems.  Fidelia strikingly improved my travel laptop/DACPort system, and I would have never paid for another expensive license just for that computer.
 
I am excited to see what comes next.

Consensus (?): if the players just deliver files to the system output and DAC then the players should sound the same.  If the players process files (upsampling, volume attenuation, equalization) then they can sound different if they use their own algorithms rather than those embedded in the OS.
 
I think someone said that it is better (potentially) to do upsampling in software rather than in hardware.  My CI Audio DAC upsamples to 24/192.  Is it better to have Fidelia upsample 16/44.1 files to 24/96 before sending them to the DAC or just send the native 16/44.1 files to the DAC?
 
I've been playing with Pure Music, Fidelia, Decibel, and others.  Those three can match the ouput sample rate to that of the files they play.  But none of them automatically changes the bit depth.  Why not???  I have to manually change the bit depth back and forth between 16 bit and 24 bit in Audio/Midi.  Is there any benefit or harm in just leaving it set at 24 bit?  
 

Bones13, nice post.