Exactly, glad, that I am not the only here one noticing this.
I'd confirm that Xine has the "crispest" sound, and I know also, what you mean by "heavier" midrange of VLC. This was my first impression, too, but now I think that VLC paints a more compact stage, or puts the audience a bit closer to the stage (still in front of it, but first row), without being louder. This would also explain, how VLC manages so well to separate and position individual instruments.
I found some music, where I like the sound of Kaffeine (or Xine) slightly more, especially, when there are Cellos or Violas, as they sound, as you correctly say, a bit heavy, or "dark", with VLC. Also, with piano concertos, it seems, that VLC seems to focus on the wooden sounding board of the piano, while the Xine family emphasizes the metal strings of the instruments, which "explains" the crisper sound.
Where the Xine family already shines, compared to the other contenders, like MPlayer, Juk etc., are voices, especially sopranos. But VLC even tops them. Not sure: Is headroom the correct English term for this? Finally, VLC is incredible precise regarding bass.
Regarding snares, they sound "brighter" with Xine, but with VLC I was able to identify individual beats more clearly, and also could "see", and their position is slightly more "defined".

These differences are very, very subtle, though, and sometimes more a question of personal likings than quality. Also, your personal ranking may depend on the music you select for your own tests. But it's definitely possible to distinguish between the following, if default settings are used in KDE 4.8.5:

VLC, the Xine bunch of players (Xine, Amarok, Kaffeine), MPlayer and SoX/play. I am unable to tell the difference between the Xine based players, though I sometimes thought Kaffeine is slightly better, while Xine is crisper, but I could not reproduce this reliably. But VLC definitely sounds different (in most cases I'd say better) compared to the other GUI players, and I could tell in 7 out of ten tests correctly, when VLC was playing, and when something else was playing. I was harder for me to distinguish between MPlayer and the Xine bunch, and virtually impossible to distinguish between the Xine players (see above).

But really, all of this is very, very subtle. All the players mentioned so far, are excellent. However, if we proceed to Audacious, I'd confirm all along, what you say. And the differences between Audacious and JuK compared to the players mentioned above are significantly less subtle. Of the two, JuK sounds more "vivid", but sometimes gives pianos wrong colors and makes concerto pianos sound like cheap E-pianos, but it's got bass and rhythm, and it's UI is perfect for quickly setting up a playlist for your next pyjama party.

gargamel

Internal filters could explain, why there are differences in sound, of course, and they could also explain, why Amarok is less loud. Does anyone know, if Amarok uses other/more/less filters internally by default than other (Xine based) players?
I've used all players with default settings. As far as I can tell, most of them sound pretty neutral, except JuK, and, to a degree, Audacious. They don't sound "bad", but sometimes, the "colours of sound" are not quite correct, I'd say.

gargamel

I see OP have very nice gear.

I'm on Sennheiser HD-25 and i must say that it matters in what source music is recorded in to hear a difference at all.

For example most soundcards sample at 48khz while file is usual 44.1khz. a normal person would'nt actual notice the difference, but playing a file which is recorded in 48khz makes a huge difference.

Perhaps, for some unknown reason vlc have something that upsamples the input file to 48khz rather then letting the soundcard doing it which results in better soundquality.

Btw! Thanks for very nice article!
and op please give me a PM if you want some real 48khz sounds (and compare them to 44khz) you WILL notice the difference.

Edit: xine sounds better, just tested myself, indeed sound in this program is more crisp and it sounds like input is better!

Yes, confirmed, I said that before. While it's relatively easy to single out VLC from other players, because it not only sounds better, but different, the differences between MPlayer and Kaffeine become evident only in certain parts of certain pieces of music.

Confirmed, again, although I wouldn't call it huge, but "easily noticeable".
A few years ago I visted a fair organised by local dealers. Not sure, if it was Naim or Linn, but one of these two U.K. legends demonstrated their then new network player with material recorded with different sampling rates. And the difference between 44.1 kHz and 96 kHz was really huge, while the difference between 96 kHz and 192 kHz was noticeable, but much more subtle.
Of course, the company had something to sell: Not only did they promote their network player, but also their newly founded record label, offering all their recordings at different sampling rates up to 192 kHz. Summary: CDs are actually crap. You better buy 192 kHz stuff

Possible, yes.

Thank you!

So you say, Xine sounds better than VLC? Or better than anything, except VLC? I have tested mostly with Xine derivatives (Amarok and Kaffeine). So, maybe I should do a few tests with pure Xine against VLC.

Cheers!

gargamel

You are talking about the differences in gain/"EQ" applied by each piece of software here.
What would help at this point is to hear from someone who knows what the path of information is from file to soundcard using each of these applications.
The subjective "this sounds better" is all well and good, and is informative, but what is needed, since there is a difference, is to find out where the errors/changes are happening.