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

[71 dB]

How much distance do you perceive? Five feet? Ten feet? Does it pop in and out as you dial the crossfeed up and down? Or is the effect consistent regardless of how much crossfeed you dial in?

I can't speak for jamesjames, but for me typically the distance is quite small, 1-2 feet, but on some recordings with very well recorded spatiality of large acoustic spaces (church music for example) the distance can be as much as 5 feets. These distances exist also without crossfeed, but the miniature soundstage is fractured and it fluctuates with the music. Hit of a piano note for example can hit on the ear, but then move further as it decays away. Crossfeed puts the minuature soundstage in order for me and the instrument stay were they are. The instruments are also more point-like sound sources and not fractured all over. For me this stability is one of the great benefits of crossfeed.

Immediately after increasing crossfeed level the miniature soundstage feels smaller, because spatial hearing reacts to the change, but adjusts back in a few seconds. The opposite happens when decreasing crossfeed level. The effect of crossfeed should always be valuated after spatial hearing has adjusted to it. Closing eyes can help "recognizing" the miniature soundstage, its size and shape.

 

[castleofargh]

It does, in fact, create an 'outside the head experience' for me. So I guess we fundamentally disagree. However, I accept you are entitled to your opinion.

But that's the thing, if different people get significantly different experiences from crossfeed, saying that it improves spatiality compared to headphones could becomes a game of sampling the listeners to cheat on statistical results. Where has it been established that it is consistently an improvement for listeners(spatially or not)?
A lot of this apparently technical discussion really comes down to personal impressions and personal priorities. At the end of the chain, the human can't be ignored.
From years of being myself a solid crossfeed user and an all time seeker of headphone saving tools, I know that most people who tried it didn't keep using it.

@71 dB argued that like EQ, most people might be rejecting crossfeed simply because they never tried it with good settings for them. I doubt it would solve everything(no I'm sure it wouldn't), but I agree that it's also part of the difficulty in assessing the actual success rate of crossfeed.
You like it, that other guy doesn't. All it shows is that people will have different experiences. Sometimes wildly different.

We understand psychoacoustics pretty well when it comes to spatial cues. But our understanding is specific to a real sound source in front of us. The moment some variables go off track, don't exist, or shouldn't exist, it becomes a gamble.
Predicting how we'll perceive a sound in front of us is made stable by how we have a lifetime of experiencing sounds that way. But something like crossfeed, who knows how a given brain will receive that half baked approximation of ... IDK, invisible speakers in an anechoic chamber but with room acoustic for all outside noises, and an interaural EQ that's not really what our head expects for that angle? That might still not be an accurate representation of the effect, but my point is, it's clearly not something we have a lifetime of experience dealing with and calibrating with the help of our eyes.

Speakers in some aspects are also acoustic monsters compared to a given sound source in front of us. But the key difference is that they do provide a consistent experience to all listeners in the same chair with still acceptable hearing. We can more easily make predictions about speakers and draw conclusions about what effect will have what impact.
Headphones don't have that consistency, and neither does crossfeed.
All that to say, those who manage to get more 3D anything from crossfeed are lucky, I never got that so I hate all of you.

 

[castleofargh]

I can't speak for jamesjames, but for me typically the distance is quite small, 1-2 feet, but on some recordings with very well recorded spatiality of large acoustic spaces (church music for example) the distance can be as much as 5 feets. These distances exist also without crossfeed, but the miniature soundstage is fractured and it fluctuates with the music. Hit of a piano note for example can hit on the ear, but then move further as it decays away. Crossfeed puts the minuature soundstage in order for me and the instrument stay were they are. The instruments are also more point-like sound sources and not fractured all over. For me this stability is one of the great benefits of crossfeed.

Immediately after increasing crossfeed level the miniature soundstage feels smaller, because spatial hearing reacts to the change, but adjusts back in a few seconds. The opposite happens when decreasing crossfeed level. The effect of crossfeed should always be valuated after spatial hearing has adjusted to it. Closing eyes can help "recognizing" the miniature soundstage, its size and shape.

That interesting. I get the maximum effect when I switch it ON. After a while, my brain somehow manages to push the instrument back to nearly 180° panning, and beside a difference in long term fatigue, I get to a point where I can't tell if crossfeed in ON or not until I switch ON/OFF again

 

[71 dB]

That interesting. I get the maximum effect when I switch it ON. After a while, my brain somehow manages to push the instrument back to nearly 180° panning, and beside a difference in long term fatigue, I get to a point where I can't tell if crossfeed in ON or not until I switch ON/OFF again

For me the panning or placement of the instrument don't change, so your experience of 180° panning on crossfeed after a while is interesting. I tried to "visualize" the miniature soundstage without crossfeed (upper) and with proper crossfeed (lower):

 

[bigshot]

I would like to hear JamesJames’ description of the effect. I have some questions that might help us figure out where this effect is coming from.

 

[jamesjames]

How much distance do you perceive? Five feet? Ten feet? Does it pop in and out as you dial the crossfeed up and down? Or is the effect consistent regardless of how much crossfeed you dial in? Could the crossfeed DSP you're using have some sort of time delay built into it?

I'm happy to answer. It's an interesting question. I've always experienced headphones as quite 'flat' in their presentation - two-dimensional. A fascinating sound, but not a particularly persuasive recreation of the sound I hear at concerts and recitals. I find with crossfeed that the presentation can be more three-dimensional - there seems to be a 'z' axis where previously there was only an 'x' and a 'y' axis to the acoustic image. It seems to create a depth perspective which was previously absent. I find the perceived effect varies significantly depending on implementation. I find crossfeed also removes a fatiguing aspect of standard headphone presentation. To date, I've not encountered digital implementations that work very well for me - although I haven't heard what I suspect might be the best (eg, dCS, Weiss). But I've found analogue implementations can be excellent to my ear - I'm thinking in particular of the Moon 430HA and SPL Phonitor amps. The effect also varies with the type of music. Smaller scale music seems to benefit most - but it's hard to predict. Of course, we're talking here about impressionistic, psychoacoustic affects, so it's hard to be precise. But with crossfeed the performance space can apparently be moved some distance from me - with a string quartet in a resonant acoustic it might be some feet. Orchestras can seem quite distant. I guess I would say that, when implemented well, crossfeed creates the impression that the listener is appropriately distant from the performance - much as you would hope to achieve with well-postioned loudspeakers. That said, it seems it's not for everyone! All I would suggest is that anyone interested in the idea should try it and make up his or her own mind.

 

[gregorio]

Crossfeed is a very coarse manipulation of spatiality and therefore looking at ILD only works fine.
When you use HRTF to create very realistic soundstage you obviously need much more detailed models and ILD alone is not enough!

You seem to be arguing against yourself here! Crossfeed IS a very course manipulation but it’s not only manipulating ILD, it crossfeeds everything, including all those factors HRTFs addresses. You can’t just ignore/dismiss those factors because crossfeed fails to handle them correctly! And, if we could simply dismiss those factors then why did anyone bother to invent HRTFs in the first place?

Same happens with speakers.

No it doesn’t, you can’t just keep repeating that.

If you play mono sound on both stereo speakers, the reality is sound is radiated from the speakers, but your brain interprets the situation as sound coming from the middle point between the speakes. Simple proof of spatiality happening in brain.

No, it’s simple proof spatiality is happening in the speakers/room and your brain is interpreting that spatiality to create its own perception.

Human spatial hearing has developed in reality so that's why what we hear and what the reality is are VERY close to each other.

Not particularly and even less so in the case of stereophonic sound.

I can't adopt a new dogma, your dogma just because you say I am wrong. I need to be convinced!

It’s not a new dogma and it’s not my dogma. HRTFs demonstrate the deficiencies of simple crossfeed, HRTFs are not new or my idea/hypothesis/dogma. So, I’ve no idea why you’re not convinced.

Increased spatiality in this context means the secondary spatial cues in the recording (there is no primary cues with headphones) are scaled into more natural form so that spatial hearing can make better sense of them.

But spatial cues are NOT scaled into more natural form, the ILD spatial cue might be but the other spatial cues are NOT, which is why spatial hearing cannot make better sense of them, although a minority of people do seem perceive that effect.

I enjoy crossfeed. I think it makes music sound more natural. Is my participation here welcomed by you?

There’s not problem at all if you enjoy crossfeed or if you have the perception the music seems more natural, because that’s in line with the science and you’re not contracting it without reliable evidence.

It does, in fact, create an 'outside the head experience' for me.

Here though we do start to potentially have a problem. Crossfeed doesn’t create an “outside the head experience” for me and many/most others, so what’s going on? Does the crossfeed know when you’re using it and create something different when I’m listening to it? Obviously not. The difference when you and I listen with crossfeed is not the crossfeed, it’s us, we have different HRTFs and different perception. So, it is not “in fact” crossfeed that creates the “outside the head experience”, it’s your perception!

G

 

[bigshot]

I'm happy to answer. It's an interesting question.

Do you mind answering specific questions? I understand your general impression. I'm trying to figure out what is creating that impression.

1) Here is a song I'd like you to listen to without crossfeed and then again with it...



How much of a difference in distance do you perceive between without and with? Five feet? Ten feet? Is it like a speaker system with a soundstage at the other side of the room from you? Or is it just a couple of inches from your face?

If each element is different, let me know which ones sound like they're inside your head and which ones sound like they're on the other side of the room (or whatever distance you perceive)... the vocals, the bongos, the piano, the woo woo's, the guitar solo. Are they all at the same distance from you, or are some things closer and some things further?

2) Do you get any perception of distance at all without crossfeed? If so, what elements sound further away?

3) Does the perception of distance pop in and out as you dial the crossfeed up and down? Is there a narrow window for the perception of distance to be apparent? Or is the effect pretty consistent regardless of how much or how little crossfeed you dial in?

4) Could the crossfeed DSP you're using have some sort of time delay built into it? I know Apple's spatial audio does. What brand and model are you using?

You don't need to write long explanations. If you could just give me direct concise answers to each question, that would be great. That will help me track it down. When you finish this, I have one other track for you to listen to and check that one out the same way... with crossfeed and without. Thanks for your helpfulness!

 

[jamesjames]

You seem to be arguing against yourself here! Crossfeed IS a very course manipulation but it’s not only manipulating ILD, it crossfeeds everything, including all those factors HRTFs addresses. You can’t just ignore/dismiss those factors because crossfeed fails to handle them correctly! And, if we could simply dismiss those factors then why did anyone bother to invent HRTFs in the first place?

No it doesn’t, you can’t just keep repeating that.

No, it’s simple proof spatiality is happening in the speakers/room and your brain is interpreting that spatiality to create its own perception.

Not particularly and even less so in the case of stereophonic sound.

It’s not a new dogma and it’s not my dogma. HRTFs demonstrate the deficiencies of simple crossfeed, HRTFs are not new or my idea/hypothesis/dogma. So, I’ve no idea why you’re not convinced.

But spatial cues are NOT scaled into more natural form, the ILD spatial cue might be but the other spatial cues are NOT, which is why spatial hearing cannot make better sense of them, although a minority of people do seem perceive that effect.

There’s not problem at all if you enjoy crossfeed or if you have the perception the music seems more natural, because that’s in line with the science and you’re not contracting it without reliable evidence.

Here though we do start to potentially have a problem. Crossfeed doesn’t create an “outside the head experience” for me and many/most others, so what’s going on? Does the crossfeed know when you’re using it and create something different when I’m listening to it? Obviously not. The difference when you and I listen with crossfeed is not the crossfeed, it’s us, we have different HRTFs and different perception. So, it is not “in fact” crossfeed that creates the “outside the head experience”, it’s your perception!

G

Hello Gregorio

I don't propose to answer the substance of this because it's completely arid.

J

 

[jamesjames]

Do you mind answering specific questions? I understand your general impression. I'm trying to figure out what is creating that impression.

1) Here is a song I'd like you to listen to without crossfeed and then again with it and tell me how much of a distance in front of you the various elements in the mix sound when you turn on the crossfeed... the vocals, the bongos, the piano, the woo woo's, the guitar solo. Are they all at the same distance from you, or are some things closer and some things further? Do you get any perception of distance without crossfeed?



2) How much of a difference in distance do you perceive between without and with? Five feet? Ten feet? Is it like a speaker system with a soundstage at the other side of the room from you? Or is it just a couple of inches from your face? If each element is different, let me know which ones sound like they're inside your head and which ones sound like they're on the other side of the room (or whatever distance you perceive).

3) Does the perception of distance pop in and out as you dial the crossfeed up and down? Is there a narrow window for the perception of distance to be apparent? Or is the effect pretty consistent regardless of how much or how little crossfeed you dial in?

4) Could the crossfeed DSP you're using have some sort of time delay built into it? I know Apple's spatial audio does. What brand and model are you using?

You don't need to write long answers. If you could just give me direct concise answers to each question, that would be great. That will help me track it down. When you finish this, I have one other track for you to listen to and check that one out the same way... with crossfeed and without. Thanks for your helpfulness!

Hello Bigshot

Re 1, I haven't listened to this, but I can say that various elements of the mix can apparently 'move' - some back, some forward, some sideways. Some bad recordings sound even worse. But I've never heard the various elements fall apart so that some are 'inside' and some 'outside' the head as you say.

Re 2, I can't add to my earlier answers.

Re 3, the parameters I've described in my earlier posts work as discussed to affect different aspects of the adjustable crossfeed 'matrix' (as described by SPL). I suggest you take a quick look at the SPL Phonitor xe manual on line to find a detailed discussion of the design and intended operation of these. The iFi analogue implemention in the iCan is, I think, different again. Meier digital approaches as found in foobar2000 and Meier amps seem to vary in levels of adjustability - depending on implementation.

Re 4, the SPL parameters just mentioned are designed to adjust for interaural time and level differences w/o DSP.

The Moon crossfeed can be switched in and out but can't be adjusted. I've tried to copy a link to an article by Tyll Hertsens on the Moon below. Among other things, it spends some time on the history of crossfeed and includes one or two of Hertsens' thoughts on the subject.

https://wilbert.nl/images/stories/v.../moon review/MOON-Neo-430HA-innerfidelity.pdf

 

[bigshot]

JamesJames... would you PLEASE listen to the track both with and without crossfeed and answer the questions. Your answers don't mean anything without being based on a specific recording. This isn't a trick. I'm trying to eliminate variables to focus on what we are trying to figure out. (It's a good song too.) You don't have to explain why you hear what you hear. I'm just trying to understand specifically *what* you are hearing. So just answer the questions directly.

I checked the manufacturer's specs on your headphone amp. The crossfeed is just a simple blending of channels. Pretty basic. No fancy effects. You say it isn't adjustable? Are most cross feeds hard wired like that? Weird. When you answer the questions, you can skip number 3 and 4 since yours isn't adjustable and doesn't involve any time delay.

The song I linked to contains several different kinds of sounds that are clearly differentiated, so it will be useful to hear how all the different elements in the song sound to you.

 

[jamesjames]

JamesJames... would you PLEASE listen to the track both with and without crossfeed and answer the questions. Your answers don't mean anything without being based on a specific recording. This isn't a trick. I'm trying to eliminate variables to focus on what we are trying to figure out. (It's a good song too.) You don't have to explain why you hear what you hear. I'm just trying to understand specifically *what* you are hearing. So just answer the questions directly.

I checked the manufacturer's specs on your headphone amp. The crossfeed is just a simple blending of channels. Pretty basic. No fancy effects. You say it isn't adjustable? Are most cross feeds hard wired like that? Weird. When you answer the questions, you can skip number 3 and 4 since yours isn't adjustable and doesn't involve any time delay.

The song I linked to contains several different kinds of sounds that are clearly differentiated, so it will be useful to hear how all the different elements in the song sound to you.

Ok! But, having listened, I'm not sure you'll find it very helpful ... you might be better reading the Hertsens article and the comments of others out there who use and like crossfeed.

So, without crossfeed, I'm afraid this sounds like a highly compressed pop song, with heavy, centered emphasis on vocals and lead guitar. The 'wall of noise' effect tends to swallow up the percussion, etc, as the levels rise. The soundstage is essentially flat. I have no sense of any performance space surrounding the performers. Actually, it sounds to me like a collection of instruments and sounds taken from a mixing desk in a studio.

With crossfeed it's essentially the same - with the wall of noise perhaps shifted back slightly (or perhaps slightly more reverberant - hard to say). I don't hear any added depth of performance space or movement between instruments.

But I do like the song - and I guess basic 'fatigue' levels would be probably lower with crossfeed over time.

I've said earlier that I listen only to classical music, and that my experience may not be so relevant for other kinds of music. I guess that the prevalence of 'field' recording of classical music is pretty important here. Even studio recordings of solo piano recitals, for example, are mic'd in a similar way. This means the recording - usually hi-res these days - is likely to present a performance space as part of the acoustic image. It's also important that the recording doesn't involve compression.

 

[jamesjames]

I meant to add that I don't know whether most crossfeed implementations allow adjustmentment. The Moon doesn't; the iFi, SPL and Meier amps do to my knowledge. I can't recall whether foobar does. I believe dCS and Weiss amp/DACs are adjustable. I seem to recall the Burson allows a couple of adjustments (unlike the SPL, for example, which is highly adjustable).

 

[bigshot]

That actually is helpful. I'll get you different kind of track tomorrow to check out with crossfeed.

 

[jamesjames]

That actually is helpful. I'll get you different kind of track tomorrow to check out with crossfeed.

Well, I'm happy to have been of service. But I'm afraid the jamesjames testing laboratory is closing its doors for good - due to competing priorities ... I don't think I have anything useful to add for the time being. And I think testing crossfeed (or other playback features) by reference to YouTube feeds probably won't be terribly helpful in allowing people to take their thinking further. As I said earlier, I think there's more than enough information out there already, prepared by people with much greater technical expertise than me. I really would suggest downloading foobar2000, or finding a store with a demo crossfeed amp, if anyone's interested in exploring. Speaking for myself, I was initially very sceptical about the idea - to the point of not trying the Moon circuit for some years. Hearing it with music and phones that I knew and liked was the important thing once I had an interest.