Well, ABX'd a few things today... -aps and -aps -m s (forces true stereo as opposed to joint stereo), and then -k. (My bad; -Y turns on the lowpass filter, but with -aps, it's already enabled) The same settings and tracks were used as in my first test, near the top of this thread.
Between joint stereo and true stereo, for the most part, I couldn't tell the difference. However, there were a few small parts in some songs (Drive (Incubus), Lithium (Nirvana), and 1969 (The Vines)) where I could.
Drive: at the beginning of the song, there's an acoustic guitar playing mainly in the right channel, and what I presume to be breath noises, very faint, in the left channel. This was more pronounced with true stereo.
Lithium: Right when the guitar is kicking into distortion, and Kurt is singing "'Cause I found God...", there's a fair bit of stereo seperation. Again, with true stereo, it was more pronounced.
1969: At the beginning, there's just a drumset playing. Again, as the others, stereo seperation of the different drums and cymbals being hit was greater with true stereo.
I also decided to try out Pink Floyd's Interstellar Overdrive in joint and true stereo. For those of you haven't heard it, about 8:40 in the song, there's some really wild panning that starts up. It sounds really awesome with headphones. Anyway, I figured it would really torture joint stereo, and I was right. It just wasn't as deep as true stereo.
Does it matter? Not really. I really had to listen to hear any differences. If you want to save about .5-1MB per song, go ahead, use joint stereo.
Now, for the -k option. I used -aps -m s (to avoid having the slightly reduced stereo seperation bias me), and decided to keep Interstellar Overdrive in, as I thought some of the synth noises might sound different with out the lowpass filter.
Abba: I could only tell near the beginning of the song (~30 seconds), where there's a very soft muted guitar picking. There was more detail in the one encoded with -k.
Alice in Chains: At the beginning of the song, where there's a tapping on the hi-hat, there was more detail and sustain in the one encoded with -k. Once the music picked up, I couldn't tell much of a difference.
Incubus: Since this song is almost entirely acoustic guitar (some cymbals occasionally), it was very difficult to make out any differences. There was a crash cymbal where I was able to identify the -k encoded file. Again, more detail, more sustain.
Pink Floyd: Again, cymbals gave it away. There's just more detail on them with -k. None of the guitar work or synth sounds made a difference with -k on, however.
Nirvana: The cymbals at the beginning were easy to pick out, as always. However, the distorted guitar mentioned above (that's only in the right channel) was more detailed with
-k on. Not much, but there was a difference.
The Vines: There was slightly more detail in the drums overall with -k on. Nothing really worth noting.
Conclusions: I would turn on -k if you're concerned about audio quality, or more importantly, if you're going to ever be listening in a fairly quiet enviroment, where these details will be heard. It just adds that last 1% to the music. As for joint stereo and true stereo, make your own decisions. I'll have it turned on just because I'm that way, but it really doesn't make much difference at all. Think of as the difference between a 192 and 256KBPS file. Only during some particular passagescontaining high amounts of stereo imaging will you notice it. And keep in mind it does add another .5-1MB or so (for average song lengths, of course) per file. -k adds a hardly noticeable amount; about .1-2MB in my experience.
Enjoy.
(-:Stephonovich