Some additions to Pio's excellent points, all of which I'm not in 100% agreement with.
The playback pathway (usually DirectSound, WaveMapper, ASIO or KernelStream) can influence the sound quality depending on what sound card (and sometimes which drivers) are used along with it.
I have tried RME DIGI 96/8 PAD with very high quality playback gear (dedicated headphone amp, shielded cables/low capacitance, HD600+Cardas cable) using both DirectSound and ASIO in Foobar.
DirectSound had horrible artifacts, which appeared to be more than just resampling effects. It's as if the noise immunity of the card was lost totally. I can't explain it, but the sound is filled with huge background noise AND rasps, scratchy sounds and random errors. To my non-professional ear they sounded more than just resampling artifacts.
ASIO on that card is faultless. WaveOut also appears to be without major artifacts (I have not listened it critically for long).
Also, be advised that using DirectSound along with resampling plugin in the player may NOT avoid resampling after the resampling plugin. This is due to the DirectSound implementation, where even same sample rate (48 kHz) as the mixer works at can get re-convereted to 48 kHz. If this happens, the clock deviation between the input stream and the mixer becomes part of the noise floor on the re-resampled bitstream.
You can test if your system is susceptible to this by outputting the bitstream of your up-to-48kHz-resampled plugin output to hard disk.
Then you need to output the same stream to the digital output of your sound card and record it again and put it onto disk as a file.
Then do a compare between these tracks to see possible errors (theoretically it's also possible that the digital input may introduce these errors, but in my experience this is more unusual).
Regardless of all the above, I agree with Pio that the other parts of your playback stream (soundcard, amplifier, impedance matching, headphones) usually have an order of magnitude greater impact on the sound quality. Then again, if you have a fluke situation like I had with my RME/Foobar combo, then the software can also impact the quality quite a lot.
As a side note, I should add that most of the so called superiority of Foobar2K over whatever-else (usu. WinAmp) appears to be software based penis-envy, with no ABX/ABC testing (or even bit-level bitstream comparisons) backing up these opinions.
I find this oddly peculiar, as most of this F2k advocacy comes from a group of so highly in favour of ABX/ABC as most HydrogenAudio members seem to be
But then again, I'm just a simple man and not a guru on these matters and I've been shown to be wrong on various occasions. So reader beware