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

[Strangelove424]

Cross-feeders do spatialize sound depending on the incoming channel separation and Meier is no exception, but that's not dynamic cross feed. Even Meier has a constant cross feed level.

The difference of Meier (a "H-topology" cross-feeder) and Linkwitz-Cmoy (a "X-topology" cross-feeder) is that Meier distributes sound according to the channel difference while Linkwitz-Cmoy emphasizes 30° angles simulating loudspeaker listening. Meier gives more vivid/aggressive/wide sound than Linkwitz-Cmoy which is more calm and relaxed.

My understanding of dynamic is that it fluctuates depending on variables in the signal, either frequency or amplitude. What do you mean by dynamic?

I enjoy Meier crossfeed because it is the most subtle I have used. Just enough to make some albums less annoying or fatiguing. It maintains the spaciousness and L/R separation I believe is still faithful to the mastering while making it palatable on headphones.

DSP is a great way to do cross-feed if you can. It's just that often you just can't use one (Spotify?) so I do all my cross-feed with my DIY cross-feed headphone adapter which is available at home no matter what the source is (CD, DVD, Blu-ray, Spotify, Youtube, TV,…) For portable music I pre-crossfeed the music before exporting them to mp3-files* for my portable player (I use a Nyquist-plugin I wrote for Audacity).

* In my opinion mp3s are "good enough" at bit rate 192 kbps or more outdoor in the noisy environment.

I like DSP because it gives complete control over range, I can set the crossover strength from 0-100 (usually have it ≤20 and mix it with many other DSPs). I like the ability to change at whim and listen on different setups, so would personally feel uncomfortable re-encoding all my files permanently with a certain DSP setting. I think it's an interesting idea, I just wouldn't do it personally. I've struggled for a while to find plugins that worked in a playback environment so I would never have to deal with rendering files in order to apply DSPs. Everything I have (including VSTs I got to work in Foobar finally) function in real time, and don't require render. Agree about mp3. I use Spotify in low quality setting when on the go. I never saw the purpose of trying to attain audio purity outdoors.

 

[Strangelove424]

It's basically a bunch of hard-panned beeps and tones, I guess the most normal name for it is 'noise music': ... he uses little to no spatialization in his mixes, either, making any notion of naturalistic listening sort of beside the point... although I realize it's an extreme case

Well, this Ryoki Ikeda is certainly not as bad as Merzbow. Anyone ever hear Merzbow? I'm not going to put a link down, but if you are curious I will warn you: turn down your volume! I found the existence of Merzbow after browsing the DR database one day and organizing results by album with least dynamic range. Merzbow and a couple other "noise artists" earned a prestigious 0 DR. Yep, that's 0 db of range. Nada. Nill. Nichts. It'll drill your flippin' brain out. Japanese electronic music seems to be heavily influenced by noise.

 

[pinnahertz]

That was the longest line by line reply I’ve ever seen. I hope I live long enough to actually getting around to reading it someday.

You could just ignore it. That's what I do with posts I find aggravating. Most of the time, anyway.

 

[pinnahertz]

My understanding of dynamic is that it fluctuates depending on variables in the signal, either frequency or amplitude. What do you mean by dynamic?

I can't find anything that says the Meier crossfeed changes dynamically. Looks like a user-adjustable setting, but signal doesn't change it. Did you find something that indicates its signal dependant?

 

[Strangelove424]

From Meier's website:

"Especially in the high frequency range, the delayed crossfeed signal interferes with the original input and attenuates specific frequencies. The frequency-curve is no longer flat but shows a larger number of dips. This is the so-called Comb-filter effect.

A unique feature of the crossfeed circuitry on the CORDA headphone amplifiers is that it "recognizes" the virtual positions of the instruments and singers in a recording. The sound of an instrument in the middle of the soundstage will be equally present in both audio-channels and isn't given any crossfeed. A crossfeed signal is only generated for instruments that are not placed at the center. The more off-center the instrument is placed, the stronger the crossfeed and the longer its delay. The frequency-curve is flat again and the Comb-filter effect is eliminated. This is called "natural crossfeed"."

...

"The original (standard) version is based on the small resistor-capacitor-network shown in the figure to the right. It can be easily recognized that the left channel input signal will also be seen at the right channel output and vice versa. A mono signal will pass unaltered and without any delay to both outputs. This version is found on CORDA amplifiers designed/built till around 2005. Crossfeed only is given for signal components with frequencies upto 1 kHz.

With crossfeed activated lower frequency signals are no longer present in one channel only but are now more evenly distributed over both channels. They are less isolated and become a more integrated part of the soundstage. They no longer stand out and this may feel as if the energy in the frequency range below 1 kHz is slightly reduced. From 2005 till now a second slighly modified filter is used that automatically compensates for this apparent bass losss.

Psychoacoustic studies have shown that our sense of direction is mainly determined by the sonic components with frequencies upto 2 kHz. However, with a simple passive network natural delay times only can be achieved for frequencies upto 1 kHz. In recent years therefore a technically more sophisticated filter was designed that allows crossfeed with appropriate delay times with signal components upto 3 kHz. This extended crossfeed filter can be found in the CLASSIC and in the (now discontinued) STAGEDAC."

 

[pinnahertz]

From Meier's website:

"Especially in the high frequency range, the delayed crossfeed signal interferes with the original input and attenuates specific frequencies. The frequency-curve is no longer flat but shows a larger number of dips. This is the so-called Comb-filter effect.

A unique feature of the crossfeed circuitry on the CORDA headphone amplifiers is that it "recognizes" the virtual positions of the instruments and singers in a recording. The sound of an instrument in the middle of the soundstage will be equally present in both audio-channels and isn't given any crossfeed. A crossfeed signal is only generated for instruments that are not placed at the center. The more off-center the instrument is placed, the stronger the crossfeed and the longer its delay. The frequency-curve is flat again and the Comb-filter effect is eliminated. This is called "natural crossfeed"."

...

"The original (standard) version is based on the small resistor-capacitor-network shown in the figure to the right. It can be easily recognized that the left channel input signal will also be seen at the right channel output and vice versa. A mono signal will pass unaltered and without any delay to both outputs. This version is found on CORDA amplifiers designed/built till around 2005. Crossfeed only is given for signal components with frequencies upto 1 kHz.

With crossfeed activated lower frequency signals are no longer present in one channel only but are now more evenly distributed over both channels. They are less isolated and become a more integrated part of the soundstage. They no longer stand out and this may feel as if the energy in the frequency range below 1 kHz is slightly reduced. From 2005 till now a second slighly modified filter is used that automatically compensates for this apparent bass losss.

Psychoacoustic studies have shown that our sense of direction is mainly determined by the sonic components with frequencies upto 2 kHz. However, with a simple passive network natural delay times only can be achieved for frequencies upto 1 kHz. In recent years therefore a technically more sophisticated filter was designed that allows crossfeed with appropriate delay times with signal components upto 3 kHz. This extended crossfeed filter can be found in the CLASSIC and in the (now discontinued) STAGEDAC."

Yes, I read that. None of that indicates a dynamic change, though. The fact that crossfeed changes depending on relative position is still a static function. In fact, my experiments fundamentally did that by extracting a crossfeed signal by developing an L-R signal then summing it with L (2L-R), then inverting it and summing it with R (2R-L). Inserting frequency response modifiers and time delay in the base L-R signal forms the approximation of head diffraction, and yes, the result is a comb filter of sorts. But it's a fixed algorithm, there is no dynamic change.

I played with dynamic change by adding variable gain control in the L-R because I found that widely separated material became over-compensated. That helped, but didn't work reliably for everything. The same process can be modified and applied to speaker acoustic crosstalk cancellation with similarly variable results. There were several products marketed that did this in the early 1980s. They only sort of worked because with speakers the phantom center image is confused by both ears hearing both speakers with two ITDs. The basis for the Carver Sonic Holography device was to compensate for that condition. It worked better, but the effect was much more fragile and affected by acoustics. It didn't apply well to headphones because in headphones you only hear one transducer per ear.

None of the above helped the middle of the head image issue, though.

 

[71 dB]

The task, as I see it, is to get the in-head, hard left-right perspective of headphones back into a more natural presentation, in essence, a more acceptable, if artificial, acoustic space.

If you record music in real space or create "synthetic" sound image using advanced effects, it is likely to contain spatial cues to achieve this. Spatial distortion often messes it up in headphone listening, but we have the solution for that: Cross-feed. If your music doesn't contain proper spatial cues, it will cause this "in-head" left-right perspective.

You mentioned it, I quoted you. You said it was easy, it's not. You should know that.

Huh?

No, it can't. Both ears still hear both speakers, even in an anechoic chamber.

Of course, but the cross-talk is minimized to the point of appearing unnatural because most people aren't used to anechoic chambers.

Even with speakers and as much crosstalk cancellation as you can manage, it's till a completely different perspective than headphones.

Yes, but that doesn't mean you can't make headphones sound exactly like loudspeakers. Ever heard of HRTF?

You brought it up and make misleading statements. I know the difference.

I brought it up? Huh?

This is one of the things that jumped out at me when I looked up the circuit. The "delay" caused by the filters is not actually time delay, it's phase shift which looks like delay when you look at one group of frequencies, but is not time delay. That time delay could be simulated well enough with an all-pass network, but not with a single pole filter. Sorry, I tried that 35 years ago. It sort of works, but not well. That's why I was disappointed. You need a real DSP to do that well.

You're right. You need a real DSP to do it well, but the phase shift it constant enough in the relevant frequency range to make it do what cross-feed is supposed to do, remove spatial distortion. At frequencies where the delay starts to fall of, our hearing moves from delay to amplitude mode in spatial hearing so it doesn't matter much. For me it works very well and I will never go back to un-cross-fed sound.

I don't know what you expected 35 years ago. I suppose you need some serious DSP to turn some badly produced 70's rock albums lacking all important spatial cues into brilliant sonic experiences… …I'm into King Crimson so I know. I ignore the sonic crappiness and concentrate on the brilliant music. Having no spatial distortion helps a lot.

At best that thing is an improvement, but it's not really doing what needs to be done. The fact that certain recordings work better than others should tell you that. Mid-head localization should not depend only on the recording, proper correction would place it outside the head all the time. That's not what you have there.

Well, I take that improvement! Of course I could have better, HRTF DSP stuff but I don't have. That stuff is for millionaires. For the money, cross-feed gives insane improvement.

Change "removes" to "reduces", and we're good. That circuit can't remove spacial distortion. I can't even minimize it.
In some cases I would agree, but certainly not all. As I referred to earlier, there is material that while mixed on speakers was happily embraced on headphones as a new, if hyper-stereo, experience. Remember, mixes are checked on headphones, especially today in contemporary popular music, since that's the market, but mixed on speakers, because mixing on speakers translates to a pleasing headphone experience, but not the other way 'round.

Remove, reduce, whatever. In my experience at some point the spatial distortion really disappears when the sound is "mono enough", is masked by the music. So that's why I say remove. If I hear spatial distortion, I increase the level of cross-feed until I don't hear it anymore.

To be completely fair, I appreciate your opinion, but do not share it.

You do have some weird opinions and I don't know what to think of you. It's as if you want to disagree no matter what I say. I try to educate people about spatial distortion and you mess things up making everything 1000 more complex for other readers. The concept of spatial distortion is complex enough as it is. There is no need to talk about how first order filters don't create "real" delays. It doesn't matter! The phase shift works, spatial distortion is reduced/removed = problem solved = enjoyable listening experience.

Speed and degree are program determined and variable. The benefit is more consistent results, the down side is more consistent results. It's just different. I didn't develope the idea any farther because the problem was the algorithm that determined the required crossfeed. It turned out it's not just the amount, that was easy to quantify, but to work well it needed delay, and the amount changes with program. What morphed out of this was abandoning the idea in favor of an idea akin to the Smyth Realizer, but I didn't have DSP in those days, so I moved on.

That's why I keep my cross feeder simple. It works and does what I want.

Actually, those efforts were very successful. Stereo headphones were fairly new, and a program featuring cool headphone mixes was a revenue source for broadcast. BTW, you're expressing opinion again. Thanks, but it's not fact, just opinion. That's why there's an off switch on cross feed!

I don't know these broadcasts so hard to tell my opinion. I started listening to radio around 1987 (and I live in Finland). It is my experience that dynamic compression on radio broadcasts reduce channel separation, hence reducing spatial distortion. I don't hear very strong spatial distortion on radio. Some, but not much.

Well, it wasn't childhood, but...
Your view is very rigid, very black and white. If you want stereo done "right" then the only way you'll be satisfied is with binaural recordings made with mics in your own ears. That works very well, but just for you.

Recording and reproduction, especially in two-channel stereo, is very much a subjective art. As you grow older (ok, sorry, just a return jab), you may realize there are lots of "rights" and "grays" in...well, everything. And there are some absolute rights and wrongs. Experience helps us to understand the difference.

Your statements show an understanding gap. The generalizations are disturbing too, like the comment about the 50s and 60s, like it was all huge ping-pong ball stereo. It wasn't, there are some very fine recordings from that time period, some even made with more that two recording channels so the phantom center could be brought under control. Most of the stereo mic techniques we still use were introduced then. And even earlier, Bell Labs research into stereophony (that doesn't mean two channels, BTW), showed that the real reproduction with accurate spacial reproduction would require a grid of over 1000 microphones and recording channels, and a speaker grid to match. The reduced the channel count until it was practical, and landed at the lower limit of 3. That was the 1930s. Give history some credit!

Huh? I can't have my favorite music as binaural recording! I must live with what is possible. Having DIY headphone adapters as cross-feeders is possible. So that's what I have. I am 46 now. How old must I be to get rights and wrongs?

The history of stereo recordings is beyond the point of cross-feed, the topic here. Of course good old recordings exists, but ping pong was "the big thing" to lure people into stereo sound.

Understanding gap? I am the first to admit that. I know enough to know how little I know. How about you?

No, spacial distortion results from the way signals are transducer.

The way recording are done creates signals that causes spatial distortion in the brain, but the recordings themselves from a technical point of view are distortion-free. Can't you even try to understand what I say? As a Finn my English might be less than perfect, but I think it's not that bad… ..God...

Yeah, right, except you are not educating people with the whole story. You've been rather definitive with your precepts, and I'm just pointing out that a few things are not so definitive. And your definition of what is "right" includes a half-baked attempt at crossfeed that doesn't take real time delay into consideration, nor the actual response curve of sound diffracting around a head, nor any thought of the angle of the phantom transducers. That's not definitive, don't portray it as final. It might not even be desirable!

I'm happy to spent my time elsewhere if my education is not welcome. I don't get one penny out of this so people can suffer spatial distortion if they want. It seems it was a mistake to join this forum if this is the mentality over here. I am talking to people who don't know much about spatial distortion, not to besser-wissers who build dynamic cross-feeders 35 years ago. Tell me how to achieve the things you mention. What does it cost? $5000? My cross-feeder is $50. Bang for the buck!

What if the creator wanted it over head? How would you know? This is again another strong effort to categorize something that is far more subjective.

Ok. Over head = ceiling loudspeakers? Atmos sound or similar...

Well, I didn't do that, but I think you just did. "Crappy" could be your opinion, and reproducing a whip-panned guitar in headphones actually was the intent of the creators some times. If you cross-feed that out, you've taken away their intention. Is that the right thing to do?
Again...perhaps yes...perhaps no. There's no way I can agree that crossfeed of the type you've defined is universally miraculous. Just as i can't agree that all album rock in the 1970s is badly produced. Not that you'll care or be impressed, but my recording background is in classical music, not 70s rock.

What if I don't like what they intended? Then I don't listen to it at all. I'm sure the same goes for you. Sorry, if you think my cross-feeder sucks. I feel bad about it. I thought I would feel good on this forum among people who understand me. I was mistaken. Life sucks, but at least I have properly cross-fed music to enjoy.

I'd probably look for a crossfeed DSP plugin and use whatever headphone amp you have. At least then the crossfeed wouldn't be limited to a simple filter and whatever phase shift it creates. It could include head diffraction, and time delay, and have all variables that are actually required. In truth why I believe you don't find much crossfeed on commercial headphone amps. It's too complex to do well.

Do what you wish. Headphone amps lack cross-feed because people hasn't been educated about spatial distortion enough. I'm trying to change that, but you ruin it. Thanks a lot man.

What we have here is a difference in opinion. I respect that you love the Linkwitz crossfeed circuit, please respect that I feel it to be inadequate. You feel crossfeed is essential and miraculous, I feel it is an occasional improvement in the implementation you've cited. I also know from experience that there is normal stereo material that crossfeed would ruin in terms of what the creators intended. I understand that from experiencing the culture of the era. Correcting that would be inauthentic, just as trying to massage stereo out of a mono recording would also be inauthentic.

I thought you were against cross-feed in general, but you are against Linkwitz? Of course one can use different cross-feeders if Linkwitz is not your cup of tea. Yes, I feel crossfeed is essential and miraculous. It revolutionized my headphone listening. Some recordings work best without cross-feed, but most of them benefit form it and I don't believe spatial distortion is intended in over 99.9 % of all music. If spatial distortion is intended then loudspeakers ruin it.

This was so frustrating...

 

[71 dB]

None of the above helped the middle of the head image issue, though.

Crossfeed doesn't even try to fix that. You need floor reflections and other spatial cues to achive that. However, if the recording itself has good spatial information, it helps when spatial distortion is removed with cross-feed.

 

[Strangelove424]

Yes, I read that. None of that indicates a dynamic change, though. The fact that crossfeed changes depending on relative position is still a static function. In fact, my experiments fundamentally did that by extracting a crossfeed signal by developing an L-R signal then summing it with L (2L-R), then inverting it and summing it with R (2R-L). Inserting frequency response modifiers and time delay in the base L-R signal forms the approximation of head diffraction, and yes, the result is a comb filter of sorts. But it's a fixed algorithm, there is no dynamic change.

It also changes depending upon frequency, but maybe I'm not following your concept of static/dynamic. An algorithm is always fixed, it's the data or the variables that are changing, and that's what allows an algorithm to react dynamically. All the algorithm needs to account for are the relationships. I'm confused as to how you would go about creating a dynamic system without setting up relationships with an algorithm. There must have been some variable that your gain control responded to.

 

[pinnahertz]

(reply to 71 dB's lengthy post...)

Well. Wow.

I think you've just pegged the superciliometer with that one.

 

[pinnahertz]

Crossfeed doesn't even try to fix that. You need floor reflections and other spatial cues to achive that. However, if the recording itself has good spatial information, it helps when spatial distortion is removed with cross-feed.

Yup.

 

[pinnahertz]

It also changes depending upon frequency, but maybe I'm not following your concept of static/dynamic. An algorithm is always fixed, it's the data or the variables that are changing, and that's what allows an algorithm to react dynamically. All the algorithm needs to account for are the relationships. I'm confused as to how you would go about creating a dynamic system without setting up relationships with an algorithm. There must have been some variable that your gain control responded to.

Static would be the specific amount of crossfeed maintains a fixed relationship to the input signal. Dynamic would be the specific amount of crossfeed is altered by some characteristic of the input signal.

I already described how I did it with a variable gain element.

 

[pinnahertz]

I thought you were against cross-feed in general, but you are against Linkwitz? Of course one can use different cross-feeders if Linkwitz is not your cup of tea. Yes, I feel crossfeed is essential and miraculous. It revolutionized my headphone listening. Some recordings work best without cross-feed, but most of them benefit form it and I don't believe spatial distortion is intended in over 99.9 % of all music. If spatial distortion is intended then loudspeakers ruin it.

I believe the LInkwitz method is fatally flawed, inadequate, crude, and isn't at all miraculous. I don't think all music benefits from that sort of crossfeed, and I don't think it is essential. If it were, it would be included in every DMP.

This was so frustrating...

Finally something we agree on.

 

[71 dB]

My understanding of dynamic is that it fluctuates depending on variables in the signal, either frequency or amplitude. What do you mean by dynamic?

Dynamic would mean the resistors and capacitors were dynamic and a function of channel separation. If you input Left only, you output left and right so that left > right. That tells you the cross feed level

level = 20*log10 (R/L). That level is constant.

I enjoy Meier crossfeed because it is the most subtle I have used. Just enough to make some albums less annoying or fatiguing. It maintains the spaciousness and L/R separation I believe is still faithful to the mastering while making it palatable on headphones.

Meier's crossfeed is good in my opinion. Nice if you enjoy.

I like DSP because it gives complete control over range, I can set the crossover strength from 0-100 (usually have it ≤20 and mix it with many other DSPs). I like the ability to change at whim and listen on different setups, so would personally feel uncomfortable re-encoding all my files permanently with a certain DSP setting. I think it's an interesting idea, I just wouldn't do it personally. I've struggled for a while to find plugins that worked in a playback environment so I would never have to deal with rendering files in order to apply DSPs. Everything I have (including VSTs I got to work in Foobar finally) function in real time, and don't require render. Agree about mp3. I use Spotify in low quality setting when on the go. I never saw the purpose of trying to attain audio purity outdoors.

I use mac so I don't have as much choice. Never used foobar. Vox player has cross feeders.

 

[71 dB]

I believe the LInkwitz method is fatally flawed, inadequate, crude, and isn't at all miraculous. I don't think all music benefits from that sort of crossfeed, and I don't think it is essential. If it were, it would be included in every DMP.

Finally something we agree on.

Linkwitz is really "easy" to implement into a headphone adapter and having multiple cross-feed levels is easy. Your claims of it being "fatally flawed" is beyond me, but each to their own. I am not promoting "Linkwitz" only. I am promoting cross-feed generally. It just happens that Linkwitz is practical in headphone adapters. I also have "line level" Meier between pre out and main in of my amp, but it's one level only.

Commercial products are often "stupid", because consumers aren't that wise.