This is the latest trend - to use asynchronous USB, where the computer is slaved to the DAC, and where various apodising filters are used to remove so-called pre-ringing in the digital signal.
The manufacturers have to think of something ... Today, dCS, Wavelength and Ayre offer apodising filter solutions, as well as Meridian, which has put this in their updated CD-player, which Stereophile drooled over as delivering "the best measured performance ever."
It's then a question of whether you can hear the added performance, or whether this is a measured benchmark that is irrelevant. And there are issues with removing pre-ringing (an artefact of anticipating the incoming digital signal), you can risk enhancing post-ringing, but that is often masked by the signal proper.
Modern electronics have jitter-performance down to picoseconds. A picosecond is 1 TRILLIONTH of a second.
0.000 000 000 001 seconds.
Desperately looking for reasons to get audiophiles to spend money on digital circuits, the industry is turning jitter into a major problem, when it isn't. What is the problem? Well, you could risk your digital circuitry either not reading the 1 or 0 right - chances of that happening are non-existent with today's equipment, unless you have actual hardware failure.
So then it's down to whether jitter can introduce timing errors in the playback of the signal - i.e. that you get good old vinyl wow and flutter, but in the digital domain, and whether this can contribute to "listening fatigue" and a sense of "digital harshness."
So let's try to get this number in perspective. A Redbook CD-signal has 44.100 samples per second, with a data-depth of 16 bits.
If we divide the circumference of the Earth at the Equator by 44.100, we end up with segments that are about 900 metres long. Let's say that you want each segment to be exactly the same length, and that there is something called jitter that is making this difficult. Even with the worst jitter performance shown above (with the power supply creating jitter of 2800 picoseconds), the individual segments would vary by at most a couple of millimetres. That's pretty good for something that is 900 meters long. And with jitter down to where we usually see it, in good (not extraordinary) components, the variation would be around 1/10 of a millimetre for a 900 metres long segment, of which there are 44.100 produced every second.
If one thinks that's a significant factor affecting listening enjoyment, then do choose extravagant jitter suppression. The greatest jitter problem in audio is the nervousness of the traditional audio industry, as the consequences of the digital music shift become clear.
The CASH page of Computeraudiophile.com is a good reference of value for money in digital audio: