To crossfeed or not to crossfeed? That is the question...

[pinnahertz]

So, my crossfeeding is wrong because of your nostalgia for hard panned stereo sound some half a century ago?

That is NOT what I said! Read it again.

You can't tell nostalgia apart from real fidelity?

Real fidelity? What's this, another of your new made-up terms? I do know what "fidelity" means. I also know that in the absolute, it doesn't exist. Every recording is a compromise, the best ones are compromises that represent the intent well.

Nothing wrong with nostalgia, I feel it too, but admit the irrationality of it. Crossfeed allows you to hear your old favorite albums more like they should have always been and perhaps even discover some artistically rich music that doesn't rely on excessive spatial effects but on real musicality and creativity.

Well, I guess the God of Cross-Feed has spoken...again.

Anyway, Who's Next album was a nice experience for me not familiar with their music previously and I liked it with crossfeed. However, crossfeed off it sounds pretty horrible imo, spatial distortion is over 75 % it seems. Typical for the genre and era.

We differ. But I challenge you to state clearly the algorithm with which you use to quantify the objective measurement of "spatial distortion" over 75%. If you state a number, there certainly MUST be a means of measurement. Or is this something only the God of Cross-Feed knows, knowledge he retains for himself? It's a number. That means there's a measurement, or an opinion. Which is it? If it's a measurement, then anyone can duplicate it and arrive at the same number. If it's an opinion, then expect it to be disagreed with.

Who's Next sounds better to me without cross-feed. Much better. It is my preference, but others my like other options. And you should research the group, listen to some of their other work, and other music of the era. It was a very formative and inventive time. And a LOT of people were discovering headphones for the first time!

 

[Zapp_Fan]

Every recording is a compromise, the best ones are compromises that represent the intent well.

I think this pretty much sums it up. The idea of objectively higher fidelity to the original stereo image as recorded is more or less a farce in my opinion. Not least because stereo mixes are often at least partly, if not entirely artificial. Live sound is a 4-dimensional object - a 3D, smooth distribution of compressions and rarefactions in air that changes over time. We're trying to experience this room-filling, ever-changing phenomenon by sampling a handful of isolated points within the room, messing with the 2-dimensional output (a mono recording), taking several of those and cramming them into two 2-dimensional containers (dual channel audio files) and then adding crossfeed, or not, at the very end, just before jamming back through another set of tranducers with innumerable problems of their own.

My point is simply that there is so much distortion between the musician and the listener, crossfeed or not, that any notion of fidelity to a real stereo image (read: 3d distribution of sound in space) is like arguing about the photorealism of a portrait drawn in charcoal. Sure, you can capture a lot of the essence of the original with good technique, and you can argue a lot about proper perspective, the right paper to use, and so forth... but it's still a black and white drawing of a living, breathing person.

 

[Strangelove424]

I think this pretty much sums it up. The idea of objectively higher fidelity to the original stereo image as recorded is more or less a farce in my opinion. Not least because stereo mixes are often at least partly, if not entirely artificial. Live sound is a 4-dimensional object - a 3D, smooth distribution of compressions and rarefactions in air that changes over time. We're trying to experience this room-filling, ever-changing phenomenon by sampling a handful of isolated points within the room, messing with the 2-dimensional output (a mono recording), taking several of those and cramming them into two 2-dimensional containers (dual channel audio files) and then adding crossfeed, or not, at the very end, just before jamming back through another set of tranducers with innumerable problems of their own.

My point is simply that there is so much distortion between the musician and the listener, crossfeed or not, that any notion of fidelity to a real stereo image (read: 3d distribution of sound in space) is like arguing about the photorealism of a portrait drawn in charcoal. Sure, you can capture a lot of the essence of the original with good technique, and you can argue a lot about proper perspective, the right paper to use, and so forth... but it's still a black and white drawing of a living, breathing person.

Your conception of stereo as dual mono, or as two flat containers, is an oversimplification. Nor does it really describe my experience of stereo, atleast on my own equipment. Since we are borrowing from visual contexts here, a 3-D image is created from stereoscopic vision, differences in perspective between left and right eyes. 3-D sound likewise comes from differences in L/R perception or L/R differences the engineer baked into the mix. The depth and spaciousness is partially a byproduct of those two 'flat' perspectives being interpreted through the brain as something with depth. Throw 3+ extra channels in for 5.1, and you have an even more amazing illusion of dimension all around. I agree that reproduction has its limitations, and cannot match a live performance, but to compare a room full of speakers to a charcoal drawing I think is a bit much. Do charcoal drawings pan?

 

[bigshot]

Mixes are definitely not real sound. They’re optimized for clarity.

 

[71 dB]

But I challenge you to state clearly the algorithm with which you use to quantify the objective measurement of "spatial distortion" over 75%. If you state a number, there certainly MUST be a means of measurement. Or is this something only the God of Cross-Feed knows, knowledge he retains for himself? It's a number. That means there's a measurement, or an opinion. Which is it? If it's a measurement, then anyone can duplicate it and arrive at the same number. If it's an opinion, then expect it to be disagreed with.

At least I have an algorithm to be challenged. It was first based on subjectivity. We find the optimal crossfeed level x and then calculate the distortion value SD from that:

SD = 100 * 10 ^ ( x / 10 ) %.​

To make things more objective, I have been testing filtering octave bands and calculating channel difference values D for them

D = S / (S + M),​

where S = abs(L-R) and M = abs(L+R)

D = 1 => L and R are identical but out of phase (antimono)
D = 0 => mono sound
D = 0.5 => S and M are equal

Now, there is a threshold value of D' which represents the largest ILD of "human hearing space". At bass it's about -5 dB. So if for example R = 10^(-5/20) * L = 0.56*L, we get:

M' = abs(L+R) = 1+0.56 = 1.56
S' = abs(L-R) = 1-0.56 = 0.44
D' = S' / (S'+M') = 0.44 / (1.56+0.44) = 0.22

Let's say I analyse a track and get D = 0.45. We want to crossfeed this so that it becomes 0.22. What is the crossfeed level x? It's a bit tricky mathematically, but the formula turns out to be:

x = 20*log10 ((D-D')/(D+D'-2*D*D')),​

So we get x = 20*log10 ((0.45-0.22)/(0.45+0.22-2*0.45*0.22)) = 20*log10 (0.23/0.472) = -6.2 dB and the corresponding spatial distortion is 100*10^(-0.62) = 24 %.

This is pretty objective, because it comes from HRTF-measurements. It is not perfect and it's work in progress. These calculations don't use ITD-information at all. That's why subjective method is more reliable imo.

Who's Next sounds better to me without cross-feed. Much better. It is my preference, but others my like other options. And you should research the group, listen to some of their other work, and other music of the era. It was a very formative and inventive time. And a LOT of people were discovering headphones for the first time!

Tangerine Dream, King Crimson, Herbie Hancock, Miles Davis and Carly Simon are my favorites of the era. Genesis and Pink Floyd don't do much for me. The 70's is a tiny fraction of all music history, so only a small fraction of attention is focused into it. If you want inventive time period in music check out electric dance music 1988-1993.

The Who is better without cross-feed? Sorry, your preferences are messed up.[/QUOTE]

 

[Zapp_Fan]

Your conception of stereo as dual mono, or as two flat containers, is an oversimplification. Nor does it really describe my experience of stereo, atleast on my own equipment. Since we are borrowing from visual contexts here, a 3-D image is created from stereoscopic vision, differences in perspective between left and right eyes. 3-D sound likewise comes from differences in L/R perception or L/R differences the engineer baked into the mix. The depth and spaciousness is partially a byproduct of those two 'flat' perspectives being interpreted through the brain as something with depth. Throw 3+ extra channels in for 5.1, and you have an even more amazing illusion of dimension all around. I agree that reproduction has its limitations, and cannot match a live performance, but to compare a room full of speakers to a charcoal drawing I think is a bit much. Do charcoal drawings pan?

OK, charcoal drawing is over-dramatic, but your analogy of a 3D movie is more accurate - but it's somewhere between something shot with 2 cameras, and something reconstructed from a set of 2D images. But still, comparing a 3D movie to the "fidelity" of being in a room with the actors is similar to what we're talking about here. Sure, you can draw comparisons, but at the end of the day, they are two entirely different experiences... 5.1 / Atmos is maybe analagous to a VR headset, I guess?

To be fair, stereo is actually dual mono in most modern formats, it just so happens that the two signals are highly correlated in certain ways

 

[pinnahertz]

At least I have an algorithm to be challenged. It was first based on subjectivity. We find the optimal crossfeed level x and then calculate the distortion value SD from that:

SD = 100 * 10 ^ ( x / 10 ) %.​

The term SD is subjective, and has not been evaluated.

To make things more objective, I have been testing filtering octave bands and calculating channel difference values D for them

D = S / (S + M),​

where S = abs(L-R) and M = abs(L+R)

D = 1 => L and R are identical but out of phase (antimono)
D = 0 => mono sound
D = 0.5 => S and M are equal

Now, there is a threshold value of D' which represents the largest ILD of "human hearing space". At bass it's about -5 dB. So if for example R = 10^(-5/20) * L = 0.56*L, we get:

M' = abs(L+R) = 1+0.56 = 1.56
S' = abs(L-R) = 1-0.56 = 0.44
D' = S' / (S'+M') = 0.44 / (1.56+0.44) = 0.22

Let's say I analyse a track and get D = 0.45. We want to crossfeed this so that it becomes 0.22. What is the crossfeed level x? It's a bit tricky mathematically, but the formula turns out to be:

x = 20*log10 ((D-D')/(D+D'-2*D*D')),​

So we get x = 20*log10 ((0.45-0.22)/(0.45+0.22-2*0.45*0.22)) = 20*log10 (0.23/0.472) = -6.2 dB and the corresponding spatial distortion is 100*10^(-0.62) = 24 %.

This is pretty objective, because it comes from HRTF-measurements. It is not perfect and it's work in progress.

It may be based on partial HRTF measurements....but:

These calculations don't use ITD-information at all. That's why subjective method is more reliable imo.

So...then the aren't based on HRTF measurements! And as soon as you claim subjectivity, you eliminate all possibility of measurement!

The very concept of "spatial distortion" is subjective, and there have not been any values assigned. There hasn't even been any statistical analysis of subjective testing! Your algorithm includes a subjective term, and is therefore subjective. Enter even one subjective term and you no longer have a means of objectivity.

Tangerine Dream, King Crimson, Herbie Hancock, Miles Davis and Carly Simon are my favorites of the era. Genesis and Pink Floyd don't do much for me. The 70's is a tiny fraction of all music history, so only a small fraction of attention is focused into it. If you want inventive time period in music check out electric dance music 1988-1993.

I largely detest 1980s electronic dance music, with one or two notable exceptions. I find most of it unlistenable. I'm pretty sure that makes me an idiot in your view, right? The 70s is a tiny fraction of music history, and it's also a tiny fraction of what I listen to. I even choose a tiny fraction of that tiny fraction. What is your point? And have we finally found someone who doesn't like "Dark Side of the Moon?" I would think that's just loaded with lovely "spatial distortion" for you to chew on.

The Who is better without cross-feed? Sorry, your preferences are messed up.

How can someone's preferences be messed up...unless they violate laws or the rights of others?

Worse, when it comes to scientific research, you've limited yourself to a single data point: your own opinion. OK, maybe two of your friends too. That is about as unscientific as it gets. Then you proclaim yourself absolutely right, and everyone differing with you as worse than wrong, they're deaf, defective, insane, all flavors of wrong, now including "messed up". I'm sure the list of derogatory terms you've actually applied is far more extensive.

Real scientific research tries to keep biases under control and out of a position to influence data, with the intent of uncovering the truth. You have fully biased research here. Actually, the term "research" would not even apply. You have no idea what people prefer and why, and worse, you don't care, and even worse, you condemn them for having an opinion different than yours. This is a very dangerous mind set, and has a long history of disaster, both for those with that mind set and those around them. You will never uncover the truth that way.

I hope you are far removed from politics.

 

[71 dB]

Some people think crossfeed "messes" with the sound. When the crossfed signal from left channel is summed with right channel we get a new right channel, but it's not "messed up", because crossfed right channel was summed with left channel too. Both channels change under the same crossfeed principles, and our hearing is able to decode what happened to the sound in the crossfeeding comparing the new left and right channels. It helps that crossfeeders simulate what happens in reality so the decoding is easy for our brain, much much easier than decoding the un-crossfed signals with excessive spatial information. People assume things out of ignorance, lack of understanding or protecting their twisted preferences. Some people think high-res audio is "better" than 16/44.1 because they lack undertanding/knowledge of digital audio and the limits of human hearing and audio reproduction technology. Similarly some people think crossfeed hardly ever is beneficial, because they lack understanding of spatial hearing.

Crossfeed messing up sound is a misunderstanding of human hearing. The sound without crossfeed is the messes up one! Well, in 98 % of cases that is...

To be fair, stereo is actually dual mono in most modern formats, it just so happens that the two signals are highly correlated in certain ways

That is very true. Spatial distortion is avoided, if the correlation is strong enough and comforms with the "human hearing space."

 

[Strangelove424]

OK, charcoal drawing is over-dramatic, but your analogy of a 3D movie is more accurate - but it's somewhere between something shot with 2 cameras, and something reconstructed from a set of 2D images. But still, comparing a 3D movie to the "fidelity" of being in a room with the actors is similar to what we're talking about here. Sure, you can draw comparisons, but at the end of the day, they are two entirely different experiences... 5.1 / Atmos is maybe analagous to a VR headset, I guess?

Yes, long before there was stereo audio, there were stereoscopes which people would place postcards into with slightly offset perspectives, which turned to 3D when peered into. "Stereo" generally refers to our biological capability to perceive depth using pairs of sensory data, and the qualities of reproducing sensory experiences that way, with depth. I would equate 5.1 to a virtualized experience as well. It's not interactive like VR is, but it can create a convincing environment all around the listener.

To be fair, stereo is actually dual mono in most modern formats, it just so happens that the two signals are highly correlated in certain ways

They are mainly recorded with mono mics and mixed stereo. I like it that way, personally. I think binaural mics sound weird the same way 3d movies look weird. Maybe that's just different strokes for different folks though.

 

[71 dB]

The term SD is subjective, and has not been evaluated.
It may be based on partial HRTF measurements....but:
So...then the aren't based on HRTF measurements! And as soon as you claim subjectivity, you eliminate all possibility of measurement!

The very concept of "spatial distortion" is subjective, and there have not been any values assigned. There hasn't even been any statistical analysis of subjective testing! Your algorithm includes a subjective term, and is therefore subjective. Enter even one subjective term and you no longer have a means of objectivity.
I largely detest 1980s electronic dance music, with one or two notable exceptions. I find most of it unlistenable. I'm pretty sure that makes me an idiot in your view, right? The 70s is a tiny fraction of music history, and it's also a tiny fraction of what I listen to. I even choose a tiny fraction of that tiny fraction. What is your point? And have we finally found someone who doesn't like "Dark Side of the Moon?" I would think that's just loaded with lovely "spatial distortion" for you to chew on.

How can someone's preferences be messed up...unless they violate laws or the rights of others?

Worse, when it comes to scientific research, you've limited yourself to a single data point: your own opinion. OK, maybe two of your friends too. That is about as unscientific as it gets. Then you proclaim yourself absolutely right, and everyone differing with you as worse than wrong, they're deaf, defective, insane, all flavors of wrong, now including "messed up". I'm sure the list of derogatory terms you've actually applied is far more extensive.

Real scientific research tries to keep biases under control and out of a position to influence data, with the intent of uncovering the truth. You have fully biased research here. Actually, the term "research" would not even apply. You have no idea what people prefer and why, and worse, you don't care, and even worse, you condemn them for having an opinion different than yours. This is a very dangerous mind set, and has a long history of disaster, both for those with that mind set and those around them. You will never uncover the truth that way.

I hope you are far removed from politics.

I am the only one here presenting ANY kind of calculations and I admit they are far from perfect. How do you justify your opinion of The Who sounding best without crossfeed? I have at least some more or less clumsy calculations to demonstrate and justify why crossfeed is needed. Spatiality is damn complex issue and I try hard to understand more everyday. Are you trying, or are you just happy with the nostalgia of hard panned stereo? How do you explain your claims that feeding excessive spatial information into your ears generate 3-dimensional sound image similar to speakers? Speakers give you natural spatial information within "human hearing space" while headphones without crossfeed give you unnatural spatial information outside "human hearing space". How could these two give you more similar results than crossfeed which scales the spatial information into similar information space to the speakers? Youe claims don't make sense to me and you have a lot of explaining to do to convert me. I try hard giving my justifications for my claims and if they are not enough for you then they aren't.

 

[pinnahertz]

I am the only one here presenting ANY kind of calculations and I admit they are far from perfect. How do you justify your opinion of The Who sounding best without crossfeed?

Once clearly identified as opinion justification becomes unnecessary, unless one wishes to convince someone else to adopt the same opinion.

I have at least some more or less clumsy calculations to demonstrate and justify why crossfeed is needed.

No, you don't! You have clumsy calculations that included your own made-up subjective parameter. You have opinion even with calculations. Your calculations do not prove need or benefit.

Spatiality is damn complex issue and I try hard to understand more everyday. Are you trying, or are you just happy with the nostalgia of hard panned stereo?

I'm trying to understand why someone would prefer cross-feed for a claimed 98% of all stereo recordings ever made. My only resource is to listen to cross-feed, which I have done extensively.

How do you explain your claims that feeding excessive spatial information into your ears generate 3-dimensional sound image similar to speakers?

I don't have to. That's not what I claimed. I do not believe speakers present much in the way of 3 dimensional sound without extensive processing.

Speakers give you natural spatial information within "human hearing space" while headphones without crossfeed give you unnatural spatial information outside "human hearing space".

The spatial presentation of speakers is unique to speakers. They don't present anything at all like a 3D space. Headphones have that potential, but that potential is also impractical. Headphones with cross-feed present another unique perspective, that of headphones with cross-feed. The effect is not identical to speakers, though it may in certain circumstances be more similar in some aspects than non-cross-feed. However, non-cross-feed headphones also present a unique perspective. None of these perspectives represents reality. All are compromises. A good compromise will convey the core sense of the original creator. In my opinion I find that the unique non-cross-feed headphone perspective retained more of what the speaker perspective had in terms of the musical mix and balance than the cross-feed version, and added a strong immersive quality that I find adds to the entertainment of the recording. I don't need to justify this, it's my subjective opinion. Everyone here knows you don't agree. You don't need to keep firing away at my one opinion. My point here is you need to get more opinions...like a massive number of them...collected in a bias-controlled means.

How could these two give you more similar results than crossfeed which scales the spatial information into similar information space to the speakers?

Because cross-feed changed the mix. It changed the balance between instruments and sound sources. It reduced distinction between instruments, it reduced depth of the mix, flattening it, and all of that reduced my enjoyment below that of my speakers.

Youe claims don't make sense to me and you have a lot of explaining to do to convert me.

I do not intend to convert you, or to make sense to you.

I try hard giving my justifications for my claims and if they are not enough for you then they aren't.

The reason you have to justify your claims and I don't is that you present yours as "right", as "fact", coss-feed as mandatory for correct listening for 98% of all recordings. That kind of claim demands justification and proof. You've provided neither. You've provided your calculations that include your own subjective variable. Proof would be a statistically significant number of listeners in a group that prefer cross-feed.

I, on the other hand, present my opinion. Not as "right" or "fact" or as a means to demand anyone listen in any particular way. I've expressed preference after testing, and I'm one guy, not statistically significant at all. I'm not trying to convince anyone of anything EXCEPT for the fact that you are not presenting fact, you're presenting OPINION, of which I have a difference of opinion. That shows that we have a 1:1 ratio, my opinion against yours. It shows there's room beyond your cross-feed edicts for other means of enjoyable headphone listening. Your edict that all headphone listening is wrong without cross-feed is unsubstantiated, it lacks any support data taken from scientific testing. It's your OPINION...presented as IMMUTABLE FACT!

As I've said several times before in this thread...that's my only problem here. I'm not anti-cross-feed, or anti-71 dB. I appreciate the work you've done, and I'm still auditioning cross-feed myself. I'm simply identifying your emphatic statements as OPINION, not fact, and trying to show that it may be possible for listeners to not prefer cross-feed for 98% of all stereo recordings on earth. I'm trying to understand why you could possibly be so definitive of cross-feed, and I'm most puzzled by your need to elevate yourself above all others, and denigrate all others, while doing so.

But the burden of proof is on you. Re-stating your opinions hasn't gotten you anywhere, has it? Prove, with research, properly done, that the average listener prefers cross-feed. Heck, that's too hard. Show in a group big enough for statistical significance, what percentage prefers cross-feed. You cannot supply that proof with equations with your own made up terms. You cannot supply that proof by reiterating your opinions, no matter how much emphasis you apply. It can only be supplied by actual research. THAT is what is now required: do the work. But until you do, I will continue to challenge your pseudo-facts, blanket claims, and statements than anyone who disagrees with you is deficient in any manner, or that you are in some what auditorily superior to the rest of the unwashed masses.

 

[71 dB]

My calculations are based on HRTF-measurements. Measurements aren't opinions. It's not an opinion that someone is 6' tall, but a measured fact. My discovery of crossfeed was a result of suddenly realizing how headphone listening easily causes excessive spatial information and when I tried crossfeed I confirmed what the scientific knowledge of human hearing tells us. There are hard scientific facts behind my "opinions." I call it understanding rather than opinions. My claims may not be 100 % accurate and correct, but they are hopefully quite correct and in time it gets more refined.

How about other people? How many percent of population even know the basic principles of human hearing? Heck, most people don't even understand decibels! If you are used to crap you may think the crap is gold. Assume ignorance and stupidity. You can't ask people, because most of the them are clueless. They need to be educated. It's sad that in our global capitalism wisdom isn't valued much. You are a valued customer as long as you have money and you blindly buy what you are brainwashed to buy. Making money selling Fidget Spinners is easier so that's what people are sold. I can only dream about a spatially cultured word where instead of Fidget Spinners all people used crossfeed to fully enjoy their music on headphones.

Asking research data to back up that people want/need crossfeed is like asking children whether or not math should be teached at school. I wonder how that would turn out...

 

[mindbomb]

This mirrors the debate between virtual surround sound and stereo in gaming. Except in that case, the deck is stacked even more against stereo, since more ambiguous sounds lead to a competitive disadvantage, and there clearly is a 3d environment that could be better represented on headphones. And yet it is still extremely difficult to get people to switch. It's classic baby duck syndrome.

For music, crossfeeds seem like the least disruptive way to address the spatial problems of stereo recordings. The alternatives are that content creators start releasing more binaural content, which are specifically made for headphone users. Or headphones have to adopt more open form factors - like the akg k1000 perhaps, or the bose soundwear.

 

[ironmine]

I agree with 71 dB. Once I found out the crossfeed technology (more than 10 years ago) and started using it, there is no way back.

I have tried many crossfeed plugins (basically, almost all of them out there, except two or three), both Foobar components and VST plugins, I have a bunch of presets for them and can invoke them for instant comparison. Meier, Isone, Redline Monitor. Before listening to an album critically, I try briefly saved presets to see which one sounds better with this particular album. I also try no crossfeed. The variant "no crossfeed" is never a winner. And never even the 2nd place. Actually, it's always the worst and the most horrible sounding.

The crossfeed technology is a major breakthrough in headphone listening. It's a revolution in headphone listening experience. But most audiophiles are Know Nothings, they are literally the most ignorant people on Earth when it comes to the understanding of how sound works. This is the good description of what a typical audiophile is: http://www.kenrockwell.com/audio/audiophile.htm

71 dB, which crossfeed do you use?

 

[castleofargh]

I agree with 71 dB. Once I found out the crossfeed technology (more than 10 years ago) and started using it, there is no way back.

I have tried many crossfeed plugins (basically, almost all of them out there, except two or three), both Foobar components and VST plugins, I have a bunch of presets for them and can invoke them for instant comparison. Meier, Isone, Redline Monitor. Before listening to an album critically, I try briefly saved presets to see which one sounds better with this particular album. I also try no crossfeed. The variant "no crossfeed" is never a winner. And never even the 2nd place. Actually, it's always the worst and the most horrible sounding.

The crossfeed technology is a major breakthrough in headphone listening. It's a revolution in headphone listening experience. But most audiophiles are Know Nothings, they are literally the most ignorant people on Earth when it comes to the understanding of how sound works. This is the good description of what a typical audiophile is: http://www.kenrockwell.com/audio/audiophile.htm

71 dB, which crossfeed do you use?

he shows it to you in his avatar ^_^. seems like good old Linkwitz design but it's hard to say at this size, I don't have the "enhance picture" feature they always have in TV shows to solve cases.