[castleofargh]

Why do we assume that our ears are the only receptors to sound. Other parts of the body may accept higher frequencies and process them for the brain. Back to double bling crossover studies of a reasonable statistical sample if you want to get anywhere near the truth and remove selling hype!

it's not about a maybe, we know the skin can feel ultrasound at very loud volumes. but we also know the ultrasonic content of the songs by looking at it, you can see that it's never even as loud as the audible range. so it's unlikely that it matters. and when it gets loud it's way up high and usually the result of noise shaping instead of actual music(and we better hope that we can't perceive it, else some DSD stuff would have made people cry).
to that you can add the simple fact that when mastered, the sound engineer is but another guy, and so it's not like he heard, tuned, and mixed the ultrasounds like he did the audible range. so at best he most likely applied some attenuation, or didn't even care for what happened to the ultrasonic content. so the chances that it ends up having any meaning that correlate with the music are small, and that's being optimistic.
 
I believe there is a difference between being open minded, and being concerned about useless stuff. if anything, it's the low frequencies that could be a game changer outside of the ears. I believe I heard Floyd Tool talk about how having something physically vibrating at low frequencies, even unrelated to the music content, had the effect of greatly improving the sense of realistic bass. for headphones that's something we might want to keep on looking. but ultrasounds... instruments have little ultrasound content, we don't perceive them unless very loud, and when very loud, because of how much energy they have, it's bad for our ears. so I say, away with ultrasounds!

 

[CloudeKr]

Oh come on this thread is degrading to a fight between audionerds with their barrages of logically flawed analogies and sciencenerds with practically no respect. It's the same thing with Apple products. People get that they're overpriced overmarketed and sensibily less practical but they buy them because they look good feel good and you got that special feel. Upwards from MP3, there's no difference in real quality that can be noticeably picked up, but cmon it feels good to have a dedicated kit playing music at that fidelity level. Don't ruin theists' days by saying that their gods don't exist, don't ruin audiophiles' by shoving these articles up their ass. It's true, logical, scientific, but not the least sensible and decent.

You'd expect half the people on head-fi to be like "wow youre right imma refund all my flac tracks and sell all my dedicated high fidelity players"

 

[castleofargh]

Oh come on this thread is degrading to a fight between audionerds with their barrages of logically flawed analogies and sciencenerds with practically no respect. It's the same thing with Apple products. People get that they're overpriced overmarketed and sensibily less practical but they buy them because they look good feel good and you got that special feel. Upwards from MP3, there's no difference in real quality that can be noticeably picked up, but cmon it feels good to have a dedicated kit playing music at that fidelity level. Don't ruin theists' days by saying that their gods don't exist, don't ruin audiophiles' by shoving these articles up their ass. It's true, logical, scientific, but not the least sensible and decent.

You'd expect half the people on head-fi to be like "wow youre right imma refund all my flac tracks and sell all my dedicated high fidelity players"

not sure it's all related. as you can see I'm one who couldn't care less for highres albums unless they cost the same as a 16bit version, but at the same time I find ludicrous to still buy old 16bit DACs with the cheap yet great stuff we now have. the improvement can go beyond playing highres files(if only for oversampling purpose or for a better handling of volume setting).
so I'm really not one to spit on progress. but ultrasounds... that's not progress we need, but human mutation.

 

[CloudeKr]

not sure it's all related. as you can see I'm one who couldn't care less for highres albums unless they cost the same as a 16bit version, but at the same time I find ludicrous to still buy old 16bit DACs with the cheap yet great stuff we now have. the improvement can go beyond playing highres files(if only for oversampling purpose or for a better handling of volume setting).
so I'm really not one to spit on progress. but ultrasounds... that's not progress we need, but human mutation.

I didn't bother reading through individual posts. I just picked up the preemble and also noted this guy defending this hifi with an analogy of table salt. So I'm stating that in general. Cheers.

 

[arnyk]

 
I say that, if people CHOOSE to base their worldview on what other people tell them, to the total exclusions of including their own personal experience, then they deserve what they get - to live in the world as other people imagine it to be.

 
I say that I know an excluded middle argument when I see one.

 

[KeithEmo]

Actually my words were chosen quite intentionally. When I say "obvious artifacts" I'm talking about things like when something that plays as a distinct tone, which stops and starts suddenly, on the WAV file, has an obvious whanging reverb tail on the MP3, or when a wire brush cymbal "disappears" entirely into a long hissy blur, or when there's an obvious harmonic content that turns a pure tone into something else. I'm not talking about imagining that things sound "slightly off", or about some obvious and ongoing alteration; I'm talking about, when I play something that I'm very familiar with, like Dark Side of the Moon, over the course of a half hour or hour recording, hearing one or two spots with distinct "clinkers" - which are repeatable with a given encoded version. (Since individual encoders make slightly different choices, this will be different for copies encoded at the same bit rate using different encoders, or even the same encoder with different settings.)
 
Incidentally, when comparing files at different bit rates, it is a reasonable argument (whether entirely true or not) that the differences "shouldn't" be audible because they occur outside what are "generally accepted" to be the limits of human audibility. The exact opposite situation holds for MP3 files. If you look at a spectral analysis of an MP3 file, or even at an oscilloscope image of the waveform, you will generally find that the MP3 file is very obviously different from the original - and that a lot of those differences fall squarely in the range of frequencies and distortion limits where they are clearly "audible" to humans; MP3 relies of the way our brain works to basically "hide" the huge errors "where we aren't listening" - and, considering how much damage is actually involved, does a remarkably good job. The problems arise for two reasons. First, because the model of what things will and will not be noticeable varies from person to person, and the encoder must use assumptions based on what "most" people will or won't notice or mind. And, second, because, since the algorithms are "purely machine based", they tend to miss "special cases" - where our brains are extra sensitive to certain types of errors in certain sounds. (Much like you probably wouldn't notice if a picture of your car had a distinct error, and showed five wheel lugs on each wheel instead of four, but, if a picture of your friends showed each with three eyes, it would be a glaring error. And, yes, the way MP3 encoding works DOES allow for errors of that magnitude - although they're supposed to be hidden where they won't be noticed.)
 
 
Quote:

   
Among the tests in the Golden Ear challenge, the frequency band and mp3 compression tests are often cited as the hardest. And the compression test doesn't go beyond 192 if memory serves. (edit: it goes up to 160 only).
 
I think "very wrong" and "obvious artifacts" are quite a strong word choices unless you've tested things blindly. As I said above, the Golden Ears test is pretty hard, and indeed I've read many people focusing in on the transients to eek (eek, mind you) out the differences at bit rates lower than 320kbps.

 

[RRod]

 

 
Yes I know how MP3 encoding works. I also know that at 320kpbs the differences are quite low in amplitude, even for your much beloved cymbals, and that even they can be impossible to ABX for many people at that high a rate (and even harder for 256 AAC). Tell me how that works out to "obvious artifacts" in the context of actually listening to music? If you've tested this stuff blind (and without touching the volume) then fine, otherwise…
 
And why do you keep bringing up different mp3 encoders? There's maybe a handful used, and their performance differences up at 320kpbs aren't much to talk about. It really just sounds like concern trolling. The question is why not use AAC or Vorbis anyway.

 

[arnyk]

  Actually my words were chosen quite intentionally. When I say "obvious artifacts" I'm talking about things like when something that plays as a distinct tone, which stops and starts suddenly, on the WAV file, has an obvious whanging reverb tail on the MP3, or when a wire brush cymbal "disappears" entirely into a long hissy blur, or when there's an obvious harmonic content that turns a pure tone into something else. I'm not talking about imagining that things sound "slightly off", or about some obvious and ongoing alteration; I'm talking about, when I play something that I'm very familiar with, like Dark Side of the Moon, over the course of a half hour or hour recording, hearing one or two spots with distinct "clinkers" - which are repeatable with a given encoded version. (Since individual encoders make slightly different choices, this will be different for copies encoded at the same bit rate using different encoders, or even the same encoder with different settings.)

 
 
Well Keith, that seems to pretty answer any any questions I might have had about whether or not you've actually done any good reliable listening tests related to any of the "Obvious artfacts" that you seem to go on and on about.  

 

[goodyfresh]

 
it's not about a maybe, we know the skin can feel ultrasound at very loud volumes. but we also know the ultrasonic content of the songs by looking at it, you can see that it's never even as loud as the audible range. so it's unlikely that it matters. and when it gets loud it's way up high and usually the result of noise shaping instead of actual music(and we better hope that we can't perceive it, else some DSD stuff would have made people cry).
to that you can add the simple fact that when mastered, the sound engineer is but another guy, and so it's not like he heard, tuned, and mixed the ultrasounds like he did the audible range. so at best he most likely applied some attenuation, or didn't even care for what happened to the ultrasonic content. so the chances that it ends up having any meaning that correlate with the music are small, and that's being optimistic.
 
I believe there is a difference between being open minded, and being concerned about useless stuff. if anything, it's the low frequencies that could be a game changer outside of the ears. I believe I heard Floyd Tool talk about how having something physically vibrating at low frequencies, even unrelated to the music content, had the effect of greatly improving the sense of realistic bass. for headphones that's something we might want to keep on looking. but ultrasounds... instruments have little ultrasound content, we don't perceive them unless very loud, and when very loud, because of how much energy they have, it's bad for our ears. so I say, away with ultrasounds!

I find it silly, as well, when people worry about such silly, useless things.

And you are right about low frequencies.  It's a known fact that infrasounds (frequencies below 20Hz) actually have profound effects on the brainwaves and perception of human beings.  Ever heard of "the hum" which is in certain areas?  Or heard of places where people get an unexplainable feeling of dread, or other types of warped perceptions (even, in extreme cases, some very mild hallucinations) due to the presence of infrasounds in the area?  It's widely thought that a good portion of ghost-sightings and reported hauntings could actually be due to hallucinations caused by infrasounds (although these certainly couldn't account for more "vivid" so-called "apparitions," whatever they might be, meh).  So it's definitely true that vibrations of our skeleton and the bones of our inner-ears, caused by frequencies below 20Hz, can greatly effect human perception.  I have a feeling that's part of why the bass sounds so visceral ("felt" as much as heard) from my Sony MDR-1A headphones, despite the fact that while certainly bass-boosted, they are not basshead headphones. . .the bass from them sounds a good deal more visceral than from many other headphones that are far more bass-boosted.  The likely reason?  The MDR-1A has some kind of aluminum-coated LCP driver tech that apparently gives it INSANE levels of sub-bass extension all the way down to 4Hz (yes, the membrane is capable of accurately reproducing sounds at the level of only four vibrations per second).  In songs that are mastered to be bass-heavy, the bass is just. . .so PHYSICAL from those headphones.

Ultrasounds, though. . .there is NO actual evidence that they have any meaningful impact on human perceptions except at EXTREMELY loud volumes.

 

[KeithEmo]

Actually
 
https://www.soundonsound.com/sos/apr12/articles/lost-in-translation.htm
 
 

   
 
Well Keith, that seems to pretty answer any any questions I might have had about whether or not you've actually done any good reliable listening tests related to any of the "Obvious artfacts" that you seem to go on and on about.  

 
You're quite right.... and I have also entirely avoided doing any proper double-blind tests to ensure that the blue pen in my pocket is really a different color than the black one (I could be imagining the difference), and I trust my own eyes that cool white bulbs really are less yellow than warm white ones. There really is a point where something becomes obvious enough that actual testing is sort of a waste of time. One minute folks seem to be claiming that the subjective differences people claim to hear between 44k files and 192k files must be all in their imagination because the differences fall outside the range of audibility (so the specs are more important than subjective experience); the next minute, they're claiming that the major, and clearly measurable, differences in MP3 files - which do fall mostly within the audible range of frequencies and amplitudes - are also meaningless (so, this time, the subjective experience is more important than the numbers)... because psychoacoustics has assured us that, even though the differences should physically be audible, they aren't subjectively audible because of masking effects.
 
(Since the whole science of psychoacoustics is really just based on statistical analysis of what people do and do not claim to hear, I'm confused as to why we should assume that the people who claim they can't hear all the measurable flaws in an MP3 file should be considered to be "credible", but the people who claim they hear differences between different sample rates should be ignored. It all starts to look to me like some people simply prefer to believe whatever "proof" agrees with what they already believe, and prefer to discount any "evidence" that proves the opposite. Could it possibly be that people only imagine that MP3 files sound OK - for the same reasons that they imagine they can hear the difference between 44k files and 192k files? or could it possibly be that they both have a point?  

)
 
Incidentally, here's a link to an article that describes, in precise detail, all of the flaws introduced by the MP3 encoding process - and what to listen for to hear them (in case anyone actually does want to listen for themselves):
 
https://www.soundonsound.com/sos/apr12/articles/lost-in-translation.htm
 
 

 

[RRod]

  (Since the whole science of psychoacoustics is really just based on statistical analysis of what people do and do not claim to hear, I'm confused as to why we should assume that the people who claim they can't hear all the measurable flaws in an MP3 file should be considered to be "credible", but the people who claim they hear differences between different sample rates should be ignored. It all starts to look to me like some people simply prefer to believe whatever "proof" agrees with what they already believe, and prefer to discount any "evidence" that proves the opposite. Could it possibly be that people only imagine that MP3 files sound OK - for the same reasons that they imagine they can hear the difference between 44k files and 192k files? or could it possibly be that they both have a point?  

)
 
Incidentally, here's a link to an article that describes, in precise detail, all of the flaws introduced by the MP3 encoding process - and what to listen for to hear them (in case anyone actually does want to listen for themselves):
 
https://www.soundonsound.com/sos/apr12/articles/lost-in-translation.htm
 
 

 
 
We've already done that ourselves, Keith, what don't you get about that? The differences between mp3 and WAV are clearly audible when *isolated*, that's trivially true. Anyone can make a difference file and verify that. The point is to hear them while music is playing at the volume where you actually listen to the music. You yourself are saying this but don't seem to listen to it.

 

[sonitus mirus]

  Actually
 
https://www.soundonsound.com/sos/apr12/articles/lost-in-translation.htm
 
 
 
You're quite right.... and I have also entirely avoided doing any proper double-blind tests to ensure that the blue pen in my pocket is really a different color than the black one (I could be imagining the difference), and I trust my own eyes that cool white bulbs really are less yellow than warm white ones. There really is a point where something becomes obvious enough that actual testing is sort of a waste of time. One minute folks seem to be claiming that the subjective differences people claim to hear between 44k files and 192k files must be all in their imagination because the differences fall outside the range of audibility (so the specs are more important than subjective experience); the next minute, they're claiming that the major, and clearly measurable, differences in MP3 files - which do fall mostly within the audible range of frequencies and amplitudes - are also meaningless (so, this time, the subjective experience is more important than the numbers)... because psychoacoustics has assured us that, even though the differences should physically be audible, they aren't subjectively audible because of masking effects.
 
(Since the whole science of psychoacoustics is really just based on statistical analysis of what people do and do not claim to hear, I'm confused as to why we should assume that the people who claim they can't hear all the measurable flaws in an MP3 file should be considered to be "credible", but the people who claim they hear differences between different sample rates should be ignored. It all starts to look to me like some people simply prefer to believe whatever "proof" agrees with what they already believe, and prefer to discount any "evidence" that proves the opposite. Could it possibly be that people only imagine that MP3 files sound OK - for the same reasons that they imagine they can hear the difference between 44k files and 192k files? or could it possibly be that they both have a point?  

)
 
Incidentally, here's a link to an article that describes, in precise detail, all of the flaws introduced by the MP3 encoding process - and what to listen for to hear them (in case anyone actually does want to listen for themselves):
 
https://www.soundonsound.com/sos/apr12/articles/lost-in-translation.htm
 
 

 
Nobody should seriously consider it to be credible evidence when someone claims they can't hear a difference between any lossy or 16-bit files and an equivalent 24-bit HD file.
 
Anyone claiming that there is an obvious difference should be able to provide some proof, as the the tools and methodology exists to accomplish this with a reasonable degree of credibility.  Trusting one's own ears is simply just piling on to the mostly worthless anecdotes.  

 

[Thad-E-Ginathom]

Oh come on this thread is degrading to a fight between audionerds with their barrages of logically flawed analogies and sciencenerds with practically no respect. ... ... ...

 
Are we supposed to respect logically-flawed analogies? Sorry, nobody told me.
 
Second best thing to do is to page down through the barrages, which I've done on the last few visits. Best thing to do is to leave this thread now.  I don't think it offers anything much any longer, other than repeated head/wall collisions.
 
<Unsubscribe>

 

[KeithEmo]

   
 
We've already done that ourselves, Keith, what don't you get about that? The differences between mp3 and WAV are clearly audible when *isolated*, that's trivially true. Anyone can make a difference file and verify that. The point is to hear them while music is playing at the volume where you actually listen to the music. You yourself are saying this but don't seem to listen to it.

 
What I don't get is how you can believe that 44k files and 192k files must logically sound the same - because no major quantifiable differences exist between them in what is agreed by most people to be the range of human hearing; yet you also seem to think that, even though there ARE major differences between MP3 files and uncompressed files that DO fall squarely inside the range of human hearing, you still think that they should also logically sound the same and the major flaws can safely be ignored.
 
To put it bluntly, I want to see as much proof that MP3 files sound the same, as the amount of proof you seem to want to see that various sample rates sound different - and for the same reasons. (The same logic that suggests that different sample rates shouldn't sound audibly different suggests that MP3 files should - assuming that we are agreeing that flaws outside the range of human hearing don't matter, and flaws that fall inside that range do matter. Since, when you analyze an MP3 file, there are significant differences inside the range of human audibility, logic suggests to me that those differences should be very audible - and any claim to the contrary seems to require "extraordinary proof"... lest we just assume that people are only imagining that they don't hear a difference. And, while I've seen plenty of tests showing that MP3 files sound "acceptably good", I see very few tests claiming to prove that there is NO audible difference. I've never heard of anyone even trying to determine the THD+N rating of an MP3 file - but I think we can all agree that it would fall AT LEAST in the full-digit percentages - since "eliminating 50% of the information" equates to 50% THD+N.)
 
I'm also willing to concede that 320k MP3 files will "sound audibly the same to most people most of the time" - but, as a matter of science, we're talking absolutes here.
 
Incidentally, just for fun, I just made up a test file with 0.5% THD - where that 0.5% THD is VERY clearly audible - even at reasonably low listening levels (so much for the claim that 0.5% THD is inaudible). If anyone wants to test that truism out for themselves, I'll be glad to post the instructions about how to do so for yourself.

 

[sonitus mirus]

  Incidentally, just for fun, I just made up a test file with 0.5% THD - where that 0.5% THD is VERY clearly audible - even at reasonably low listening levels (so much for the claim that 0.5% THD is inaudible). If anyone wants to test that truism out for themselves, I'll be glad to post the instructions about how to do so for yourself.

 
I consider .5% to be high, but you didn't bother to post any relevant details about how this measurement was obtained.
 
I have an inexpensive Peavey USB DAC that has the following THD specifications, and this is worse than anything else I have purchased in the last decade.
 
THD + Noise: < .015% @ 1 kHz, -9 dBu full scale output into 10k Ohm