[sonitus mirus]

For the OP's question, yes, I think it is worth getting high-end headphones if you only use 320 kbps audio files.  In my experience, headphones continue to be the most important part of my setup, with both lossless and 320 kbps lossy files.  All of the DBT that I have personally done indicates that I can't tell a difference with an overwhelming number of songs.  I have not seen any reliable proof to suggest otherwise.  If anyone claims to easily be able to identify a difference, they have not provided any confirmation that is accessible to the masses and this premise remains anecdotal to me, so I will continue to remain skeptical for now.
 
I would, again, strongly suggest that any listener attempt their own ABX test with their current equipment.  It's the only way to be sure. Any DBT can only prove that someone can hear a difference.  It cannot be used to scientifically prove two files sound identical.  You can only prove this to yourself.  Those claiming the differences are obvious and simple to detect without providing any proof are suspect, as this can easily be shown by posting the results of an ABX test.  

 

[EddieE]

In all of the years this discussion has been repeated endlessly, I am yet to encounter anyone who has passed a double blind test on this. Not once.

This includes multiple people try and concede that, while they really thoight they could with sighted tests, they failed the double blind test. Multiple others claiming they could and avoiding requests for evidence.

That should really be the end of it, but of course it wont be...

 

[sonitus mirus]

I've seen plenty of evidence to suggest that it can be done (check out the Hydrogen Audio forums or even peruse some papers in AES if you have access), and with some people, quite routinely.  Although, it usually has to be the right song, or even the right section of a particular song.  Even then, it almost always seems very difficult for the tester, and when I've had access to the same samples used in the ABX, it was impossible for me and several others to tell any difference between the two files.  This is true when I'm even provided with something specific to listen for between the two files.  There are "expert" listeners that have the ears and/or training to be able to reliably hear a difference.  None of this changes my opinion on high-end headphones being worth the expense if someone only uses 320 kbps lossy files.  It's simply an opinion, though, but that is the only answer the OP is going to find for now.

 

[pp312]

So what you're saying is, yes, it's possible for a few Golden Ears to hear the difference if they're really listening for it, but in the course of normal, everyday listening sessions the difference becomes irrelevant.
 
So it comes down to a question of, is it possible to hear a difference if you really try (and are trained in what to listen for), and for the ordinary person not generally listening for differences does it matter anyway? Refering back to the original question, the answer surely is, yes, absolutely, better phones will make everything sound better, be it MP3 or FLAC.  

 

[SomeGuyDude]

  I've seen plenty of evidence to suggest that it can be done (check out the Hydrogen Audio forums or even peruse some papers in AES if you have access), and with some people, quite routinely.  Although, it usually has to be the right song, or even the right section of a particular song.  Even then, it almost always seems very difficult for the tester, and when I've had access to the same samples used in the ABX, it was impossible for me and several others to tell any difference between the two files.  This is true when I'm even provided with something specific to listen for between the two files.  There are "expert" listeners that have the ears and/or training to be able to reliably hear a difference.  None of this changes my opinion on high-end headphones being worth the expense if someone only uses 320 kbps lossy files.  It's simply an opinion, though, but that is the only answer the OP is going to find for now.

 
This is exactly correct. Yes, it CAN be done, but lots of things CAN be done. 
 
There's something even more than the ABX double blind, and one that I think is more important. Which is if you give someone 320 and tell them it's lossless, will they be able to, on their own, pick up that you lied to them or otherwise say "really? This sure doesn't sound lossless..." to which the answer is "absolutely not". 
 
I would be further willing to bet that for anyone who listens to especially busy or "dirty" music (myself, for example, with a lot of metal in my collection) wouldn't even know if you went from lossless to 192 unless we were told beforehand.

 

[Dogmatrix]

   
This is exactly correct. Yes, it CAN be done, but lots of things CAN be done. 
 
There's something even more than the ABX double blind, and one that I think is more important. Which is if you give someone 320 and tell them it's lossless, will they be able to, on their own, pick up that you lied to them or otherwise say "really? This sure doesn't sound lossless..." to which the answer is "absolutely not". 
 
I would be further willing to bet that for anyone who listens to especially busy or "dirty" music (myself, for example, with a lot of metal in my collection) wouldn't even know if you went from lossless to 192 unless we were told beforehand.

Again with the ridiculous sweeping statements, this statement is incorrect unless you have conducted the test with 100% success on 100% of the population
I grow bored with the lack of debating technique being displayed here
 

  So what you're saying is, yes, it's possible for a few Golden Ears to hear the difference if they're really listening for it, but in the course of normal, everyday listening sessions the difference becomes irrelevant.
 
So it comes down to a question of, is it possible to hear a difference if you really try (and are trained in what to listen for), and for the ordinary person not generally listening for differences does it matter anyway? Refering back to the original question, the answer surely is, yes, absolutely, better phones will make everything sound better, be it MP3 or FLAC.  

This is much better, it begins to approach the point of the debate which is one of magnitude,
Is the loss of quality in compression greater than the increase in quality of high end headphones.
Or
Does the sound quality of modern high end headphones easily compensate for any loss in quality due to compression.
 
Personally I agree with the many who have pointed to the low cost availability of high volume memory which renders compression obsolete.

 

[SomeGuyDude]

  Again with the ridiculous sweeping statements, this statement is incorrect unless you have conducted the test with 100% success on 100% of the population
I grow bored with the lack of debating technique being displayed here

 
I'll tell you what. You find me a single person who could pass that test and I will happily retract everything I've ever said.
 
Actually, you find me anyone who can reliably pass a double blind test on 320 vs lossless and I'll do it. So far, such a person has not appeared. As above, once in a while you get someone who has a moderate success rate, but even at best it's only somewhat above random guessing, and even then only with a narrow band of genre listening and with someone who's specifically trained for it. 
 
So let's take that information. If people can't even tell the difference during a direct ABX test, how many do you think could tell the difference when they're not swapping back and forth? Hint: it's somewhere around zero. This is what we in the biz call "extrapolation". It's when you take known information and apply it to further hypotheses. 

 

[Dogmatrix]

   
I'll tell you what. You find me a single person who could pass that test and I will happily retract everything I've ever said.
 
Actually, you find me anyone who can reliably pass a double blind test on 320 vs lossless and I'll do it. So far, such a person has not appeared. As above, once in a while you get someone who has a moderate success rate, but even at best it's only somewhat above random guessing, and even then only with a narrow band of genre listening and with someone who's specifically trained for it. 
 
So let's take that information. If people can't even tell the difference during a direct ABX test, how many do you think could tell the difference when they're not swapping back and forth? Hint: it's somewhere around zero. This is what we in the biz call "extrapolation". It's when you take known information and apply it to further hypotheses. 

This avenue of debate is a dead end because the information you refer to is unreliable and so any extrapolation simply magnifies doubt
ABX tests contain too many variables to produce a definitive result

 

[JamesHuntington]

Just upgrade music at the same rate as setup. That or where can one put the blame on quality? It's like putting 87 octane in a race car. And then changing every bit of the car before realizing it's the gas that is the weakest link.

 

[forsberg]

I converted a FLAC file to 320k, set them up for ABX in foobar2000, couldn't tell a difference. I did the test about 20 times. Those where I got right, I was guessing. So to answer your quesiton - yes, it's worth it.

 

[Beocord]

  So it comes down to a question of, is it possible to hear a difference if you really try (and are trained in what to listen for), and for the ordinary person not generally listening for differences does it matter anyway?

 
Funny thing happened the other day. I did the philips golden ears challenge. I passed the part where you have to tell differences between different mp3 qualities easily with no mistakes. Later that night I was listening spotify premium which is supposed to be 320kbps ogg vorbis... I don't remember which song it was but something sounded kinda off so I checked if I was using the highest bitrate. For some reason it was set to 160kbps instead of 320. I have no idea how long it had been that way, maybe days could be weeks. Differences between 160kbps and 320 are clearly audible but even that can go unnoticed for quite a while if there is no reason to question the sound quality.
 
Weird thing is that I still prefer my music as high quality as possible. 24/192 dvd-a and such... There really is no point in having 5 minute songs with filesize of 300mb but since it's possible why the heck not.

 

[Dogmatrix]

Reclining in my comfy chair, basking in the majestic orange glow of the Sophia Princess in my custom built one of a kind tube amp driving my HD800, I can see the question in a new light.
If for the sake of argument we consider a live performance the ideal then a 320kps recording is some distance from this ideal. Does a high end headphone move one closer to or further away from the ideal ?
It is my experience that through the HD650 320kps and CD occupy the same distance from ideal but the HD800 reveals the flat dynamic and grainy texture of 320kps thus moving me further away.
With this in mind I have to conclude that on a diet of 320kps the HD650 is the better choice and so it is not worth getting a high-end headphone in this case.

 

[EddieE]

Funny thing happened the other day. I did the philips golden ears challenge. I passed the part where you have to tell differences between different mp3 qualities easily with no mistakes. Later that night I was listening spotify premium which is supposed to be 320kbps ogg vorbis... I don't remember which song it was but something sounded kinda off so I checked if I was using the highest bitrate. For some reason it was set to 160kbps instead of 320. I have no idea how long it had been that way, maybe days could be weeks. Differences between 160kbps and 320 are clearly audible but even that can go unnoticed for quite a while if there is no reason to question the sound quality.

Weird thing is that I still prefer my music as high quality as possible. 24/192 dvd-a and such... There really is no point in having 5 minute songs with filesize of 300mb but since it's possible why the heck not.

I totally get going with higher quality than needed for the sake of idealogical perfection. My colection is FLAC when I know I can't pass a DB test between it and well encoded, error corrected 320CBR.

I would draw the line at 24bit though, as there is, scientifically speaking, no reason at all it should sound better.

With lossless to lossy, there is a measurable and real difference. The debate is around whether it can actually be heard.

With 24bit, it is just the latest marketing push to make us re-buy music we already own and "upgrade" gear when we already have perfectly good components.

The music and audio business is constantly on a disingenuous drive to get us to "upgrade".

 

[Beocord]

Yea..I agree. Clever people at the recording industry. At one point when I got a lot of 24/192 I was amazed by the sound quality. Some bach cello stuff by Linn records is of course awesome becouse of the bitrate, not because they actually paid attention how to record that stuff. Hehhe Few rock bands that generally have awesome quality in their albums such as Dire Straits and Pink Floyd sound as good to my ears through spotify. Latest albums by Iron Maiden sound like crap even on lp. I rarely use my turntable these days though... With hegel+phonitor I finally have a system that sounds fairly analog to my ears.
 
About the need to make people upgrade... Is it impossible to "convert" a normal album into binaural? Make a perfect room for recording, spent 10 million into the gear... In theory you would have sound that is allmost as good as the real thing, if the original recording was good. Then have binaural dummy record what is played by the best audio gear available and have professional engineers fix issues that have occured. 
 
I mean...people pay for minor tweaks made into the recording. Imagine stuff like greatest hits by led zeppelin, now in binaural. I'd pay for that.

 

[pp312]

With lossless to lossy, there is a measurable and real difference. The debate is around whether it can actually be heard.

 

 
And there you have it in a nutshell. Earlier in the debate some of us got caught up on the idea that others were denying that there was in fact any difference between lossy and lossless per se. Obviously lossy loses data, so cannot be the same as lossless. The debate is therefore whether the loss can be detected. Initially we all got hung up on the question of whether it can be detected at all, by anyone under any circumstances. I considered this irrelevant and have tried to drag the question around to whether even those who can detect it under test conditions are ever really going to be bothered by it in a typical listening situation. Because that after all is all that matters. The whole recording chain is riddled with subtle losses that serve to remove us a step at a time from live sound. No one can seriously say his system reproduces a genuinely live sound. And that's from CD. When you consider the considerable step from what goes into the microphone during a recording to what we hear from our systems, is the difference between 320 and FLAC of any serious significance? I say not. Yes, in theory lossless is superior to lossy--obviously--but real-world practise (innumerable DBT tests) has shown that it has no significant affect on listening pleasure: it presents no real barrier to the music. I know the pro-lossless brigade will never accept that, but that's as I see it.
 
Incidentally, I still don't accept the 'storage space is cheap' argument. It may be, but transfering huge files between flash drives etc is deadly. I don't own an ipod, but the situation must be the same there.