Well finally we can agree on something.

Your "point" still doesn't explain the fascination with different interconnects (bullocks) and cable risers (more bullocks) and people buying speakers based on their price tag and some really inventive marketing.

I'm sorry to say that yes: you're arguing with someone who's mind is completely closed to the possibility that cables and audio risers and tiki torches strategically placed about the room while a Tibetan Monk chants in the backyard is going to increase my stereo system's audio fidelity. I've listened to a lot of stereos (and, FYI, I think it's extremely rude of you to act as though the people you're arguing must have bricks in their ears to disagree with you, of all people. I know I don't appreciate it), built a lot of stereo equipment, and have a lot of experience in this domain, and not once have I ever heard a substantial difference going from one cable to the next.

Really, can we stop discussing this ridiculous nonsense? Had I known anyone was so insecure about their audio system, I would have forgone my original comments many times over. I think fancy interconnects and cable risers are a bunch of voodoo, and yes: my mind is made up. If you're not comfortable with the fact that some random stranger on the Internet thinks you wasted your money, then I suggest therapy.



I just don't think those offer much (if anything) beyond a CD. Truthfully, I don't even listen to CDs anymore; all twenty gigs of music I listen to are on my computer, compressed to about 300 kbps in variable bit rate WMA files. This is because it's easier for me to listen to and the fidelity is quite good--I'd gladly take the taste test against the CD I tore it off of.

I wouldn't listen to LPs for no other reason than dealing with them sounds like a pain in the butt.


Truthfully, if you want my opinion, there has been zero (0) innovation in audio playback in the last forty years. This is largely because:

1. It was already damn good to begin with.
2. There is a very serious law of diminishing returns with respect to high end audio.
3. Assuming 1 and 2 are true, the only way to continue making money in the audio domain is to come out with new, flashy junk on a regular basis to entice consumers into throwing their money away.

Most of the so-called "innovation" in the audio industry is, in fact, very clever marketing, junk science, and messing with people's head.

And this is my final post. I'm sick and tired of being considered some sort of boorish moron simply because we have conflicting opinions.

post edited: no point furthering this debate, whether with actual points or insults. there is no evidence available that will satisfy both sides, without that nothing further needs to be said.

Alright, I'm going to have to pull out my yellow card right now. This is a public forum and no matter what names you decide to call each other (sticks and stones and all), it's really rude to refer to an active participant negatively in the third person. BobDog is reading this too, you know.

Personally, as much fun as it would be to see BobDog post numbers proving whatever, I don't think I'd be satisfied until I tested him myself with my own protocols. This is why I sometimes consider myself to have a scientific mind. Of course when it comes down to interpreting the data statistically, I suck.

FWIW, I do believe that there may be things we can hear that we cannot measure. I also believe that there can be phenomena that is detectable but cannot be explained easily. This follows from the fact that we are able to sense before we are able to think or analyze. This is why challenging BobDog to explain how and why things work is silly. He's already mentioned that he's not really an engineering authority. We could get similar results by challenging race car drivers how different motor oils affect engine performance. They might be able to tell the difference, but f*** if they know how it works. The question of measureability vs human detecting ability is very much debatable, and I have no problems letting this debate go forth in this topic.

Now regarding the ad hominems and insults, I'm going to quote my parents, "I don't care who started it, but it's going to stop now."

Alright now, Play ball.

EDIT: Removed quote that was taken back. My point still stands.

I know you dont care who started it but BobDog did referring to me in the third person trying to explain the theory of science. Maybe yeha took this a step further *shrug*. Despite the debate getting ugly at times I have learned a lot reading this thread.

One thing I would like to know though, is how do you guys encode your CDs? I simply use CDDA with 256 kb/s mp3. I keep hearing about EAC, but if I limit my cdrom (lets say to 12x) using nero, isnt it getting the right bits anyway? I imagine there is error checking and parity bits etc in there so it should be just fine. Anyone wanna bother explaining this to me?

not all ripping methods are created equal, it's good that you've heard of eac

this thread and included links would be a good introduction to why eac (or plextools, should it support your drive) is a better choice, and why just limiting drive reading speed can potentially make no difference at all in how many errors are missed or corrected. i found eac after multiple rippers claimed 100% accuracy on my cds, however listening to and analyzing the output revealed pops and cracks where interpolation hadn't even been attempted. burst ripping strikes again.

as for how you store the music you rip, it all depends on the threshold of your hearing with regards to compression artifacts. i would say that 256 kbps cbr mp3s are never an optimal choice - if you're using mp3, it's hopefully for hardware compatibility as many formats have now supplanted it in terms of quality per kbps (hydrogenaudio blind tests proved this). if you'd like to stay with mp3 for hardware compatibility reasons, consider encoding with the "lame" encoder's "--preset standard", it was extensively tuned by the founder of hydrogenaudio with input from the best listeners there, and should average around 180 kbps for most music. it started life as a fork of lame specifying "--alt-presets", but listening tests forced its inclusion into the lame mainline and actually became the new standard encoding profile. for the vast majority of the population this will be perceptually transparent from the original. disregard any information you see from the mp3 "scene" which says joint stereo is bad, vbr is bad, lowpasses are bad, etc. - these people know nothing about the theory behind audio compression, they just stumbled onto broken iso-derivative encoders like bladeenc, noticed the flawed output it created and thought it was the format's fault, not the buggy encoder.

if you're looking for better quality without regarding hardware compatibility, mpc is still your best bet - less problem samples than aac, much much faster to encode thanks to the simpler analysis that a subband encoder can get away with and the standard preset will average about 160 kbps. it's what i "archive" my cd collection with, since even when problem samples are encountered the lack of a transformation stage means artifacts are extremely subtle. for almost-as-high quality (again, this has been proven with blind tests) with growing hardware-compatibility aac via nero has been getting extemely good reviews.

i'm not a terribly big fan of ogg vorbis. the container format (ogg) is a nightmare - whenever monty designed it he mustn't have had much experience with media container formats, the mplayer and ffmpeg boys have been so critical of it they started their own container format (nut), and after attempting an ogg parser myself i can't blame them. vorbis as a compression format has potential, particularly the tuned variants that hydrogenaudio members have programmed, but the extremely slow development cycle, over-stretched main programmer and lack of hardware support make this format a questionable choice over mpc for anything other than sub-100 kbps encoding. if you're after low-bitrate files for streaming or previews, the new aac profiles are better, for high-bitrate transparency, ogg requires more bits than mpc and aac to reach it. right now it's just a middle-range format after it's time.

if you're ever interested in a listening tests to see how well you can discern audio compression artifacts, sign up to this newsletter for a notification of whenever roberto starts another distributed abx test. might not be for a while though, they're fairly stressing for him to administer, especially after the slashdotting that the last one got.

summary: eac secure ripping, mp3 (with lame) for convenience, mpc for absolute quality, aac for a nice balance of the two. all imho of course.

People can call me whatever they want... I don't take it personally and nothing I say is meant personally either--it's just that some of the things that have been said here are so far off base (like certain peoples' understanding of scientific method, just as a for instance.).

If you read back through my posts, you will see I answered every one of yeha's points... but it seems that he is even less interested in taking the time to listen to dissenting post than he is in listening to quality audio.

All but one: CD vs. SACD. As anyone will tell you, SACD allows for more information to be stored than CD--it is capable of "24bit resolution" and the equivalent of "96kHz sampling" in PCM (as even DAD-A's backers will admit--though they note that these numbers cannot be obtained simultaneously). More info. would generally mean better sound... but then again, yeha is the one who thinks that compression doesn't matter, toss away those extra bits, who needs 'em?, so I think tiring to drive this point home would be a lost cause. There are many other interesting issues regarding DSD vs. PCM, indeed noise shaping being amongst them, but it is too late to get into all that here.

Bottom line (for me), where I had SACDs and CDs of the same alblum, SACD sounded better most of the time than CD (some SACDs are mixed engineered abysmally) to me and everyone else I had listen to my Meitner deck--even when CDs were up-sampled to DSD. I did not AB, I did not ABX, I did not single or double blind (so if you wish to say I was brainwashed you are welcome, though that begs the question of why certain SACDs sounded so terrible to me if it was all just some mind trick...??? Anyone who wants to answer this post, please answer this point clearly: if my preference for SACD was a figment of my imagination, why did certain SACDs sound so clearly bad to ordinary--such as Norah Jones' Fly Away--or whatever it was called?)

As have I, like how Stereo Review stayed in business for so long.

I had wondered the same thing. Edward Ng was very helpful on the whole EAC thing--perhaps he will weigh in again? As I know so little about computer audio (now), I will not comment on yeha's post on this topic... he may even be correct, stranger things have happened.

Yeah, EAC isn't perfect, even in secure mode, personally i do test and copy, to assure the same checksum on both runs, and I encode in flac, overkill? Yes, but I like the warm fuzzy feeling of a perfect copy.

AtW

Exactly the same reason why I FLAC; even if I supposedly can't hear a difference, I can be 100% confident that this way, there is no difference to be heard, either way, and I have the storage space to get away with it, anyway.

There's nothing more for me to chime in with regards to EAC; it does its job, and it does it well. There are a couple good guides around the 'net I can try to find one that discusses how to configure it to rip that supposedly provides a better rip than the default settings. Don't have time right now to look for it, though.

And btw, I've been following this thread all along, post-for-post, but when it got more hostile, I decided to stay out of it. Mod's aren't supposed to pick sides or get involved in the heat, opinion or not. Anyway, I think my opinions sit in the middle ground, see as I always believe in shades of grey, particularly here.

-Ed

you did reply to most everything i said (definitely not everything), unfortunately your responses folded into this little word "nuance" which you are unable to define, and unless language is defined it's meaningless (i.e. your replies may as well not have existed at all). well here's your chance - exactly what facet of a pressure wave does nuance detail? let me guess, something that only ears can perceive that can't be picked up by any electrical device on earth, which you're unable to explain in terms of physical characteristics of a sound wave.

from my side of the fence, all i hear from most audiophiles is "electrical instruments can't detect the same nuances as my ears! your experimentally equivalent cable lacks nuance! non-elevated cables lose all nuance too! cds lack nuance but lps and sacds are packed full of the stuff! nuance nuance nuance!!"

fine, great, super. like i said earlier without mutually agreeable evidence this whole debate is pointless, and so far the evidence has only been coming from one side. no amount of discussion will change anyone's mind unless there are bulletproof tests to accompany it, and it's not easy to find good double-blind tests since many in the audiophile community shy away from it. i have a few friends currently in university, i'll see if during the summer they can talk a prof into letting us run some abx tests with the university's equipment, but i'm pretty sure such tests would be ignored.

oh, and as for the negative implications made about lossy audio compression, here's the perfect opportunity to step up to the plate - i can upload a file that you can download and burn to an audio cd to test at your leisure - the results (and you can run the test as many times with as many people as you want for all i care) will give us some actual EVIDENCE to analyze. right now, your positions on certain audio equipment has as little credibility in my eyes as my thoughts have in yours, we both realize that, and the only way past it is by performing tests. i'm up for it.

alright..



eh? the tests you just listed are completely meaningless, the only accurate way to compare the mediums would be if both cd and sacd were mastered from the same master signal (almost never the case, and extremely difficult to verify for commercial releases thanks to new copy protection), calibrated for identical playback volumes (ditto) and then compared in a double-blind fashion. the fact that none of those facets were present, and you still thought it was somehow a valid comparison between the mediums, well, i really don't know what to say. that's as far as possible from a valid comparison as you could possibly have run, except maybe comparing two completely different albums.

the idea that you can have a poorly-engineered cd, sacd or lp is hardly ground-breaking. it doesn't matter how accurately you can recreate crap, no medium is immune to garbage-in-garbage-out.

i don't know if that was as clear as you wanted it to be, but it's all i can do.

The electromechanical values of speaker will change quite radically during the first 24-72 hours of use. If you check out some diy speaker sites you are bound to find before/after measurements there. Any intelligent speaker designer will design the crossover to be used with drivers that have had their suspension loosened after some use, since these are the parameters the speakers will be having for almost their entire usable life. This also explains why speakers sound better after a while.

Another interesting thing, opamps are popular in the amp diy community right now and many seem to be observing that new opamps heat up a lot at first but the problem goes away after a while.

gee whiz, I'm gone to a wedding for two days and there's another two pages of posts....

yeha, I have done a good bit of compression codec comparing myself and although I haven't checked out EAC, in a blind test I can tell the difference between a 192kbps mp4 and CD audio 19 times out of 20 (we got really bored one day....). I know the principles behind lossy encoding (had to write a paper on them) and they make perfect sense to me...but I'm still missing something from those recordings. It's great that they work for most people, though.

@ everyone who discussed our spatial reasoning on the last page...I had a professor in college who got his doctorate in that area of study, wrote quite a few papers on it as well. He discovered all kinds of interesting things in his research, like that if a person sites perfectly still and a speaker is panned in an arc to the side (back to front in a circular room), that person will actually hear the sound as arcing above their head. It turns out that we need to move our heads to get several different readings on a sound source; if we are still our ears are much less reliable. He also found that our front-to-back reasoning is very flawed when we aren't moving; many people incorrectly identified a sound source in front of them as behind them, and vice-versa. Interestingly enough, this is true even at an angle; a sound source 30 degrees to the left of a person's nose will often be identified as a source 30 degrees to the left of the back of their head. If you're curious, google Dr. Wes Bulla or Dr. Wesley Bulla, you should be able to find some of his publications.

So lossless isn't really lossless?

He's not talking about a losslesss codec; I think he made a typo there.

-Ed

To answer the question, lossless, is lossless, but sometimes reading a cd is not lossless, a cd-reader has an offset for example, which can be around a millisecond, which you loose (unless the cd-drive can read into lead in/out). Although a ms of silence at the end isn't exactly a catastrophe.

AtW

Crap, meant "lossy", thanks for the good proofing out there. Edit made.

I'm certain I've mentioned this before, but I use PlexTools to rip to a Wav, and subsequently encode it using MPC --standard. If I didn't have a Plextor drive, I'd use EAC. EAC is technically better than PlexTools in terms of security (I believe PlexTools relies on C2 information), but I'm willing to make a small compromise for increased ripping speed.

Rumor has it (though I haven't tested) that PlexTools also makes short work of copy protection schemes too.

Ok, here's a question: If speakers/amps/etc. have to be burned in to sound their best .... why don't companies burn them in before they sell them? It's common practice in the computer industry (I think) to test a system for stability before it is sold by doing a two day burn-in, so why doesn't the audio industry do this? Even a burn-in time of 100 hours is only four days, which is hardly an inconvenience (no labour required).

So. We're stuck on "nuance", a wonderfully vague term that describes something that might be perceptible, we're not quite sure. Whatever it is, it's certainly not measurable in terms of physical laws given the state of our current understanding of how perceived sound works in physics.

I have seen some ABXing (via hydrogenaudio links) that looked pretty decent to me, although I'm not sure if BobDog, our resident statistician would approve. However, I think part of BobDog's problem with this approach is that it may not be capturing everything there is to hear, i.e. "nuance". I honestly can't say one way or the other whether all the subtleties of music can be accounted for in terms of noticeable differences (i.e. "transparency" in an ABX test). I can't shake the feeling that there's a large subconscious component to music that may not be obvious to even a careful listener. Is it possible that this "nuance" that is only perceivable over a long period of time is a subconscious emotion that cannot be directly perceived (i.e. active listening won't catch it)?

I'd like to propose a way to test for "nuance" beyond what can be found by ABX testing. I think BobDog's strongest point thus far is that listening to music is an emotional experience, not simply a transposition of physical sound waves into brain activity. So, why not devise a test to quantify the emotional response to a particular piece of music? I know there's a huge difficulty in trying to quantify an "emotional" experience, but I also know there's a long (and fairly successful) history of this kind of testing in psychological circles.

Here's what I suggest: First, we all take yeha's transparency test for audio compression. We all decide on particular piece of music at a particular bitrate that is transparent for all of us, but is close to the limit of transparency. This will give us an audio file with "missing" information that cannot be directly perceived by actively listening. Then, we take this file and the uncompressed original, and run tests on a random sample of the population, keeping the playback system constant (and of the highest quality, according to scientific measurements). Instead of playing both samples and asking if they can tell the difference, we play one of the samples, and ask them to rate their emotional response to it (1=hate the song, 10=love it). Gradually, we'd build up an average response to each sample. I'm sure BobDog could tell us how large a sample we'd need to generate a statistically significant difference. I'd imagine we'd need a very large sample, since "nuance", if there is such a thing, is likely to produce only a VERY small difference. We'd probably have to agree on a certain confidence interval, beyond which we agree that, even if a difference is detected, it's too small to matter (possibly the most difficult part of my proposal).

Using this method, one of three things could happen:
1: No statistically signficant difference is detected: Either our confidence interval is too large (There is no such thing as "nuance", or it is too small an effect to matter)
2: The original scores better than the compressed version: "Nuance" exists, there are differences in music beyond what is directly perceivable, and these differences are obliterated by lossy compression (and possibly non-audiophile stereo systems).
3: The compressed version scores better than the original: "Nuance" exists, there are differences in music beyond what is directly perceivable, and these differences are falsely added to more "accurate" recordings by slightly flaws in lossy compression (and possibly high-end audiophile systems).

No matter which result is obtained, we at least have some evidence whether the goal of reproducing music should be accuracy or merely what sounds good (and has nuance). If result 1 is obtained, this is strong evidence that differences such as cable swaps are psychological effects like the placebo effect. If results 2 or 3 are obtained, then there is evidence that the most scientifically "accurate" measurements do not properly capture what we enjoy most about music, and should be weighted less heavily, as BobDog has been suggesting.

Obviously, this will not resolve every question, but it would give both sides of this argument a lot to think about, and perhaps advance our collective understanding of music.

Devonavar, I think you bring up some good points. Ones that have possibly been touched on here, but certainly not at such length. This could be an interesting experiment. To throw in my experience on this subject.....I once had to make a study where we were testing if people could respond to any sounds outside their range of hearing. Pro audio equipment has gone way past the original 44.1kHz sample rate, which gives us an upper limit of 22.05kHz wavelength that can be reproduced. So what we had to figure out was if anything above 20kHz or below 20Hz made any impact on a listener at all. We took some movie scores that we knew had foley work dropping down to 5Hz or so, and we took a few projects that had been mixed at higher sampling rates than basic redbook audio. Using filters, we played people the recordings both with the extra 5Hz tones and without; the one with the extra information was almost unanimously voted better, although no one could exactly voice why. The most typical comment was that it just "felt bigger". Just for kicks, we filtered out everything above 20Hz and played that....no one heard or felt a thing. It was much the same with the high-frequency stuff....people thought it sounded "more open". We only went up to a 96kHz sampling rate, b/c to my knowledge there are not yet any commercially viable speakers that can reproduce a 96kHz tone (what a 192kHz sampling rate gives you). Again, we also played "just" the above-20kHz frequencies and no one heard anything.

So I think this could probably be extrapolated to say that there are also nuances within the audible frequency spectrum that also influence the way we hear things, subconciously, even though they wouldn't be able to be heard by themselves.

Say you have one instrument that plays as high as 20KHz; no problem. Say you have two instruments playing a 20KHz sound, off-phase by 180 degrees from each other. Now imagine four of them, all off-phase by different amounts. Can 44.1KHz reproduce all four instruments clearly enough for the listener to discern them from each other? How about six, seven or eight? It's not a problem for smaller groups of instrumentalists, but if you have a full-blown orchestra, per say, I think the waters muddy when sampling rate isn't sufficiently high.

Would that be sufficiently technical explaination?

How about six instruments, one playing pure tone 20KHz, one 18Khz, one 17.5KHz, one 15KHz, one 14.2KHz and one 11KHz. Will they all be clearly discernable from each other in a recording sampled at 44.1KHz? How would 96KHz sampling compare?

-Ed

As long as a simple tone at 20 khz is correctly reproduced, it should be no problem(?), no matter how many instruments there are, i's only a waveform, and any wavefrom can be reproduced by a sum of simple waves (sines).

AtW

Would you really hear all six instruments though? Particularly, if they're not absolutely pure tones (real life instruments rarely achieve absolutely pure tones; well, nothing in real life is absolutely perfect)?

-Ed

Would you want to hear six instruments all playing at 20kHz? To answer your question, though, the sampling rate need only be as high as the highest frequency that is being recorded. There's a technical explanation, I just can't remember what it is right now

As I said, any complex (no matter how many and how complex instruments create the resulting wave) waveform can be represented as a sum of sines (fourier), as long as you can reproduce pure tones up to 20 khz perfect, you can reproduce any combination of theese. And all waveforms from instruments can be represented as a sum of sines.

AtW

If it's in the recording, I want to hear it.

-Ed

I don't work with audio, but I do worth with video, and in that realm this would be a vast oversimplification.

I work for a company that manufactures IP-based video cameras. I work a lot with raw image types in both the YUV and RGB color spaces.

RGB is pretty obvious to envision--you have a Red, Green, and Blue component, and the combination of those three channels results in the desired color. If you specify eight bits for each channel, you end up with 24-bit color. Because 24 bits doesn't align on a 32 or 64 bit boundary in computer memory, each pixel is padded up to 32 bits because that's a much easier alignment to work with. Sometimes the extra eight bits are used for alpha, and sometimes they're used for nothing.

Needless to say, this amount of data for representing a pixel is a complete waste. Empirical testing shows that the human eye is much more sensitive to changes in intensity (luma) over chroma; the result of these observations was a slightly different method for storing video, called the YUV colorspace. The YUV colorspace also has three components: a Y component for luma or intensity (think of it as black and white), U for the blue chrominance, and V for the red. This site does a decent job of explaining the differences:

http://www.joemaller.com/fcp/fxscript_yuv_color.shtml

There are many different YUV-based formats, but almost all of them are based upon 12 and 16 bits per pixel. I work with a type called YV12, which has one U and V sample for every 4 Y samples--all of the data types are 8 bit, so the ultimate resolution of the video is still 24 bits, but they're not devoting as much information to chroma. Needless to say, I can't tell a difference between 24 bit RGB and 12 bit YV12, and the YV12 results in an instant halving of your data size by removing data that your eye wasn't sensitive to in the first place. Almost all MPEG2 based decoders (and all WMV decoders) prefer operation in the YUV colorspaces, because it's inherently easier to work with the colors separated out.

So yes: sometimes you CAN discard bits and end up with something better, because all storage methods are just a means of representation--with some ways of representation being much more efficient than others. In other words, there are creative ways of representing digital audio and video that don't involve throwing ridiculous amounts of storage space/resolution at the problem (which, to be honest, often creates a much larger problem).

I know this is video, but it shares quite a bit in common with audio signals. You should be open to the prospect that:

more information does not always equal better quality

A great article on the subject (love wikipedia):

http://en.wikipedia.org/wiki/YUV

edit: another one....love google too:

http://www.animemusicvideos.org/guides/ ... space.html

yes rgb is a waste - if you're going to spend 24 bits on a pixel, you can get much more perceived accuracy or equal rgb's perceived accuracy with fewer bits. digital audio is similar - those 16 bits would have been better spent on logarithmically quantized amplitudes instead of linearly. now this caught my eye..



this could be read as a digital source revealing more (new) information after a transformation, which we certainly don't want to communicate! perhaps instead of "discard bits" we could say "sample from a source using fewer bits" as you certainly don't gain anything from putting an already-digital image through a colorspace conversion

typically though in compression we hold video to much lower standards than audio - i've done a fair bit of mpeg-1/2/4 codec programming and the amount of detail you can get away with discarding is amazing, however once you know what to look for it gets quite easy to abx. with audio it's all-or-nothing - no video codec hoping for usability is going for lossy transparency (at least i've never heard of any, and can't see much need for one), however with audio that goal is attainable at tantalizingly low bitrates - sub 200 kbps for most signals.

as to the multiple instrument query, yes all frequencies the instruments create below the recording system's input filter will be recorded, however each additional instrument (assuming equal amplitude) will be recorded with a lower dynamic range, since you must attenuate the entire signal to avoid saturating your recording medium. you can record a solo saxophone player with more clarity than a duet of saxophonists each playing as loud as the first (while in both cases retaining all frequency information over a given amplitude threshold), if you're going for maximum dynamic range. that said the psychoacoustic masking involved with multiple instruments will outweigh the medium's limitations, imho.

Green Shoes - interesting test results! i hope of course it was double-blind and multi-session but i've actually never found an abx test of extremely-low frequencies. one would think they'd be felt more than heard - i know i've felt (low) sounds i couldn't hear, that rumble made my stomach feel like a case of food poisoning. the few tests i found for supersonic abx results are all offline, but the summaries showed no one was able to pass an abx test of two signals that differed only by their > 20 khz content.

for the nuance search, the only valid test of opacity is when two signals are presented and a listener can discern between them with a given confidence level - if they can't reach that level they're not hearing a difference and the signals are perceptually identical, regardless of how much they'd like to believe otherwise. having people give an untrained and largely random grade to a passage after one playing without a reference, will test people's moods on the day of the test, their musical preferences, their level of caring, whether they want to get in your pants, etc. and maybe the level of nuance contained in the signal as one of the many other things being tested. you could run the results through anova but i doubt there'd be anything useful coming from it, the noise would be extremely high as there are so many different things being compared at once. one hydrogenaudio member claimed that he had headaches after listening to lossy audio for several hours, but not with lossless audio. i believe a long-time-scale abx test was theorized but i think it fell through.

to test this nuance theory that's exactly what you'd need - a long-term abx test that had rounds of days or weeks at a time. it doesn't sound very appealing to coordinate such a test, especially with multiple participants, and would also severely limit the subject's usage of their own equipment as a lot of legitimate and common usage could destroy the blindness of whatever's being tested.