[ironmine]

That's why we low pass filter the leaked sound at 800 Hz (1st order butterworth is a common choice)

How about this, for a start? :

I take the left channel, split it into two, one goes directly to the output, the other one goes to the equalizer, before it gets mixed with the right channel.
Which EQ mode should I choose: linear phase or natural (minimum phase)?
How much should I attenuate the sound of the left channel coming from the EQ plugin before it is mixed with the right channel? -12 dB? -20 dB?

 

[castleofargh]

I already mentioned this thread https://www.head-fi.org/threads/recording-impulse-responses-for-speaker-virtualization.890719/ . you can ask there for some basic stereo impulses(or graphs from them), and see how well they align with some crossfeed options.
or you could get human measurements from the few HRTF databases available online and pick 30°(or whatever you fancy) to see what is really going on for some people.
obviously this would become really relevant with your own measurements. that could look maybe something like this:

recorded from about 30° left speaker. red is picked at left ear, blue at right ear. keep in mind that speaker, room, mic and mic calibration are all non strictly flat, so what's mainly relevant here beside some general response caused by the ear and how the headphone should probably try to cautiously aim for something similar, is the attenuation per frequency I get from the opposite ear. as that's in principle what crossfeed is aiming to replicate badly.
as for the delay, from the impulses I seem to be near 250µs on this one.
someone else would have something else. not completely different, but noticeably different anyway.

 

[ironmine]

I already mentioned this thread https://www.head-fi.org/threads/recording-impulse-responses-for-speaker-virtualization.890719/ . you can ask there for some basic stereo impulses(or graphs from them), and see how well they align with some crossfeed options.
or you could get human measurements from the few HRTF databases available online and pick 30°(or whatever you fancy) to see what is really going on for some people.
obviously this would become really relevant with your own measurements. that could look maybe something like this:

recorded from about 30° left speaker. red is picked at left ear, blue at right ear. keep in mind that speaker, room, mic and mic calibration are all non strictly flat, so what's mainly relevant here beside some general response caused by the ear and how the headphone should probably try to cautiously aim for something similar, is the attenuation per frequency I get from the opposite ear. as that's in principle what crossfeed is aiming to replicate badly.
as for the delay, from the impulses I seem to be near 250µs on this one.
someone else would have something else. not completely different, but noticeably different anyway.

I have the same (my own) measurements. The problem with these measurements is that they contain not only the effects of a head, but also, primarily, the effects of a room and speakers. I do not want to mimic the imperfections of my room and speakers. I want to transcend them.

 

[gregorio]

[1] Very little if anything can be gained. That's my problem. How to gain?
[2] I am really really bad at selling my opinions to other people as you have seen.
[3] Since recordings are mixed for speakers. [4] Crossfeed is kind of a "mapper" that takes the original spatial parameters and "maps" them into parameters that are more natural to human ear.
[4a] This is what my reason says and it's what my ears say.
[4b] That's why it is so difficult for me to believe I am wrong.
[4c] If I was wrong then the implications would be I can't trust my reason and senses at all.
[5] If I agree with Gregorio I disagree with my ears!
[6] The solutions are the there, but they are not easy.

1. Asked and answered, numerous times!

2. This isn’t the “sell your opinions” sub forum and if your opinions contradict the facts/science then they’re ALWAYS going to be impossible to sell here!

3. That’s not strictly true, as mentioned numerous times. Mixes are virtually always at least checked and approved on headphones and commonly somewhat adjusted. This of course brings us to artistic intent and your bizarre belief that your intent/preference supersedes the artists’ and should be imposed on all other consumers/listeners.

4. No it doesn’t! The “original spatial parameters” (on commercial music recordings) are not natural to start with and crossfeed doesn’t magically make them natural. How many times?
4a. Your reasoning is flawed and one of the main reasons why it’s flawed is because your ears are not “saying” anything, what you’re listening to is your perception, not your ears. Which is EXACTLY the same mistake so many audiophiles make when asserting improvements!
4b. Which is again exactly why so many audiophiles also can’t believe their false assertions are wrong!
4c. Science clearly demonstrates that you can’t trust your (perception of) your senses and therefore, as your reasoning omits the fact that different people’s perception varies, your reasoning contradicts the science, is flawed and will continue to be impossible “to sell” here!

5. Again, it’s not your ears but your perception, you do not directly hear your ears. So if you are to agree with the science (rather than contradict it) and change the word “ears” in your statement for the word “ perception”, then NO, you can both agree with me AND your personal perception (but then you can’t apply your personal perception to everyone else)!

6. The solutions ARE EASY for a rational mind. For someone with an irrational mind, who believes they are a messiah and their preferences should be imposed on everyone else, then maybe the solutions aren’t easy but I wouldn’t know, that’s an issue to discuss with a personal therapist/psychiatrist, NOT an issue for this sub forum!

G

 

[71 dB]

How about this, for a start? :

I take the left channel, split it into two, one goes directly to the output, the other one goes to the equalizer, before it gets mixed with the right channel.

1. Which EQ mode should I choose: linear phase or natural (minimum phase)?

2. How much should I attenuate the sound of the left channel coming from the EQ plugin before it is mixed with the right channel? -12 dB? -20 dB?

1. Minimum phase as in the picture. Lowpass (in the pic you have lowshelf filter), -6 dB/oct.

2. That's the crossfeed level, which as I said in my opinion should be variable between -12 dB and -1 dB (that's gain = attenuation 12 dB to 1 dB) The former is for recordings needing only the final gentle touch to be "headphone ready" and the latter for sound with very large channel separation (ping pong stereo, movie sound downmixed from surround sound). So basically every recording has it's crossfeed level setting and you are supposed to set it by ear so that it's the lowest setting taking away any excessive spatiality and makes the sound appear natural. So, if for example -7 dB does it, by -8 dB (one dB resolution is good enoug imo) not, the "proper" crossfeed level for that recording is -7 dB. When you learn to hear what crossfeed does to the sound it's easy to set this level (takes 1-3 seconds for me). If you want one fixed level for crossfeed you are having a compromise of how when that level agrees with "proper" level. Typical values for fixed crossfeeders are -10 dB … -6 dB. My first diy crossfeeder had only -8 dB setting, but I soon recogized the need for adjustable crossfeed level so I first modified the crossfeeder to have options -10 dB, -8 dB and -1 dB and soon after that being wiser I built my current 6 level crossfeeder.

In your diagram you are missing the treble boost block between input and output in the lower branch. Just put another fabfilter plugin there (highself filter, 800 Hz, Q=1) the boost should vary with the crossfeed level according to the the pic I attached in my previous post. If that's too difficult then a fixed +2 dB is a compromise.

 

[ironmine]

Just now I tried this idea:

I opened in Room EQ Wizard my room measurements which I made several months ago:

1) Left speaker sound measured near the left ear;
2) Left speaker sound measured near the right ear;
3) Right speaker sound measured near the right ear;
4) Right speaker sound measured near the left ear;

and exported them (no smoothing) as impulse files.

Then I built this monstrous VST chain in ART-Teknika. The first four instances of MConvolutionEZ convolver process the left and right channels using the impulses (1,2,3,4).
Another two instances of the convolver apply the room correction impulse* (which I normally use when I listen through speakers) and the headphone correction impulse for my Denon D2000.
The Voxengo Sound Delay plugin delays (2) and (4) by 250 ms. I can also use it to attenuate (2) and (4) but it sounds best when I don't reduce their volume.

(*I need to try removing the phase correction component from this impulse, as it now corrects both for room response and for speakers phase non-linearity.)

The resulting sound is really close to what I hear in my room through speakers. Overall, the resemblance is indeed striking. I am a bit shocked with the result.

 

[ironmine]

1. Minimum phase as in the picture. Lowpass (in the pic you have lowshelf filter), -6 dB/oct.

2. That's the crossfeed level, which as I said in my opinion should be variable between -12 dB and -1 dB (that's gain = attenuation 12 dB to 1 dB) The former is for recordings needing only the final gentle touch to be "headphone ready" and the latter for sound with very large channel separation (ping pong stereo, movie sound downmixed from surround sound). So basically every recording has it's crossfeed level setting and you are supposed to set it by ear so that it's the lowest setting taking away any excessive spatiality and makes the sound appear natural. So, if for example -7 dB does it, by -8 dB (one dB resolution is good enoug imo) not, the "proper" crossfeed level for that recording is -7 dB. When you learn to hear what crossfeed does to the sound it's easy to set this level (takes 1-3 seconds for me). If you want one fixed level for crossfeed you are having a compromise of how when that level agrees with "proper" level. Typical values for fixed crossfeeders are -10 dB … -6 dB. My first diy crossfeeder had only -8 dB setting, but I soon recogized the need for adjustable crossfeed level so I first modified the crossfeeder to have options -10 dB, -8 dB and -1 dB and soon after that being wiser I built my current 6 level crossfeeder.

In your diagram you are missing the treble boost block between input and output in the lower branch. Just put another fabfilter plugin there (highself filter, 800 Hz, Q=1) the boost should vary with the crossfeed level according to the the pic I attached in my previous post. If that's too difficult then a fixed +2 dB is a compromise.

Ok, thanks, 71dB, I will try it !

 

[71 dB]

2. This isn’t the “sell your opinions” sub forum and if your opinions contradict the facts/science then they’re ALWAYS going to be impossible to sell here!

3. That’s not strictly true, as mentioned numerous times. Mixes are virtually always at least checked and approved on headphones and commonly somewhat adjusted. This of course brings us to artistic intent and your bizarre belief that your intent/preference supersedes the artists’ and should be imposed on all other consumers/listeners.

4. No it doesn’t! The “original spatial parameters” (on commercial music recordings) are not natural to start with and crossfeed doesn’t magically make them natural. How many times?
4a. Your reasoning is flawed and one of the main reasons why it’s flawed is because your ears are not “saying” anything, what you’re listening to is your perception, not your ears. Which is EXACTLY the same mistake so many audiophiles make when asserting improvements!
4b. Which is again exactly why so many audiophiles also can’t believe their false assertions are wrong!
4c. Science clearly demonstrates that you can’t trust your (perception of) your senses and therefore, as your reasoning omits the fact that different people’s perception varies, your reasoning contradicts the science, is flawed and will continue to be impossible “to sell” here!

5. Again, it’s not your ears but your perception, you do not directly hear your ears. So if you are to agree with the science (rather than contradict it) and change the word “ears” in your statement for the word “ perception”, then NO, you can both agree with me AND your personal perception (but then you can’t apply your personal perception to everyone else)!

6. The solutions ARE EASY for a rational mind. For someone with an irrational mind, who believes they are a messiah and their preferences should be imposed on everyone else, then maybe the solutions aren’t easy but I wouldn’t know, that’s an issue to discuss with a personal therapist/psychiatrist, NOT an issue for this sub forum!

G

2. I don't think my opinions contradict facts/science. Science doesn't say artistic intent should be respected, you do. That's YOUR opinion and it's not very scientific is it? Science doesn't say who to respect. Science doesn't care about artistic intent. Science tells us how we hear and I am trying to apply that to headphone listening and for me that work in the way that science often works because it's science. It is insulting to be called unscientific.

3. You are correct, but it's not black and white. Recordings require different levels of crossfeeding depending on how well they are "checked on headphones." Modern pop music for example often respects headphone listeners a lot (mono bass etc.) and if I use crossfeed it's weak crossfeed to take the remnants of excessive spatiality away. Olders recordings? In my opinion very rarely sound "headphone ready."

4. Well, not "technically" natural, but spatial hearing can be fooled (the whole concept of stereo sound is based on fooling spatial hearing) and to our hearing the spatiality appears natural IF the spatial parameters are close enough to natural spatiality. Since the natural range for ILD at low frequencies for example is 0-3 dB, limiting the ILD of a recording from range 0-14 dB to 0-3 dB makes it sound more natural to spatial hearing.

4a. When I say my ears say something I of course mean my perceptions. I don't know what happen in my ears (well scientifically I know because I have studied human hearing). I only know what I perceive. That's what matter. So much for flawed reasoning.

4b. Usually those false assertions are related to placebo-effect etc. snake oil things.

4c. My reasoning contradicts science because it agrees with science? What? I know very well I can't trust my perception, but I can trust my reason. You think I am some kind of idiot who hasn't thought about crossfeed for one second? I have pondered these things for years! And you have the audacity to tell me my reasoning contradicts science? If it did I'm sure crossfeed would have NEVER invented or used. Why bother if it's so against science? The only thing it is against is your fetish to worship artistic intent in an idiotic way that causes excessive spatiality with headphones. It's your business if you prever headphones sound "as it is". That's your perception and I that's your choice, but don't call other people unscientific when they have scientific background and education + years of pondering these things! You have said NOTHING that debunks my reasoning. All you have is some illusive concerns about ITD and what not, but you haven't demonstrated convincingly how these debunk my reasoning. I admit crossfeed is not perfect and has these little insignificant things and that kills your concerns. As I have said million times: Nobody cares about how acoustic crossfeed (and ER + room acoustics) messes ITD with speakers, but if you dare simulating that with a crossfeed, suddenly ITD is a gigantic problem. Re-evaluate your own reasoning before attacking someone with the knowledge and understanding of these things I have. Even if you knew more than I know that doesn't mean I don't know anything. I have 7 years of crossfeed experience. How many years do you want to take me seriously? 20? 30?

5. Deal! So much of this has been semantic nonsense.

6. I am not telling anyone they are forced to use crossfeed. My point is science does support/justify crossfeed. That's doesn't mean everyone's perception agrees with science. My perception seems to agree with science, perhaps because I am a science-oriented person rather than "artistic intent-oriented". If the artistic intent sounds natural on speakers and unnatural on headphones, I name the speaker version what the artist intented and use crossfeed to have something closer to that on headphones. I also use my own head. If my perception of the spatiality with headphones (without crossfeed) is crap, I conclude it's because the artist didn't check for headphones and I do it myself for her/him using crossfeed.

 

[ironmine]

71dB, is it the correct way? :

It sounds very nice!

Any other way to improve the scheme? Maybe I should insert a Mid-Side plugin to control the volume of the mid channel separately from the sides?

The sound is a bit too close to me. How to remove it away from me so it would be in front of me but at a certain distance from me?

 

[gregorio]

2. I don't think my opinions contradict facts/science.
[2a] Science doesn't say who to respect.
[2b] Science tells us how we hear and I am trying to apply that to headphone listening.
4a. I only know what I perceive. That's what matter. So much for flawed reasoning.
4c. My reasoning contradicts science because it agrees with science? What?
[4c1] I know very well I can't trust my perception, but I can trust my reason.
[4c2] You think I am some kind of idiot who hasn't thought about crossfeed for one second?

2. It’s clear you think that, it’s also clear (to everyone except you) why and how you think that. Namely, by ignoring/dismissing those parts of science/the facts that you’re contradicting, perfect circular reasoning!! How many times?
2a. Correct, science doesn’t tell us who to respect, common sense does! Again, if I buy an album by say Bjork, whose artistic intents/preferences do I want to hear, mine, Mozart’s, Bjork’s or yours? Duh, your argument is ridiculous!
2b. Science does tell us how we hear (broadly speaking) but YOU ARE NOT applying that to headphone listening! Even by YOUR OWN admission you are only applying certain bits of science, which ISN’T science, it’s pseudoscience. Furthermore, the science you don’t apply/account for, you dismiss on the basis that you don’t/can’t perceive the negative effects. This contradicts science because they do exist, we can measure them and science demonstrates they are audible. So STOP saying science supports your opinions because it does the opposite!

4a. So much flawed reasoning indeed! Your perception is what matters to you BUT it doesn’t matter to me or to science!! Excellent example of flawed reasoning, thanks for demonstrating!!

4c. Your reasoning contradicts science because although it agrees with certain parts of science, it dismisses and contradicts the demonstrated science that perception/preferences will vary from person to person! How many times???
4c1. Which is an OBVIOUS FALLACY because you are basing your reasoning on perception you admit you “can’t trust”, duh!
4c2. No, I think you’re “some kind of idiot” for basing your reasoning on only part of the facts, ignoring and contradicting others and then trying to ram your fallacies down everyone’s throat, in a science forum of all places!!!

The rest of you post is just more of the same ridiculous nonsense, self contradictions and ego massaging. Round and round you go, bla bla bla!

G

 

[71 dB]

71dB, is it the correct way? :

Looks ok to me.

1. It sounds very nice!

2. Any other way to improve the scheme? Maybe I should insert a Mid-Side plugin to control the volume of the mid channel separately from the sides?

3. The sound is a bit too close to me. How to remove it away from me so it would be in front of me but at a certain distance from me?

1. That's good.

2. Crossfeed itself is effectively a frequency dependent mid-side plugin, but you may want to reduce the channel separation little more at high frequencies.

3. This type of crossfeed processing doesn't actively remove sound from the listener. It does it to some extend by reducing excessive spatiality that make hard panned sounds appear at your ears. How close the sound is is dependent on the recording itself. Dry recordings appear more close than recordings with heavy reverberation/spatial effects.

 

[71 dB]

2. It’s clear you think that, it’s also clear (to everyone except you) why and how you think that. Namely, by ignoring/dismissing those parts of science/the facts that you’re contradicting, perfect circular reasoning!! How many times?
2a. Correct, science doesn’t tell us who to respect, common sense does! Again, if I buy an album by say Bjork, whose artistic intents/preferences do I want to hear, mine, Mozart’s, Bjork’s or yours? Duh, your argument is ridiculous!
2b. Science does tell us how we hear (broadly speaking) but YOU ARE NOT applying that to headphone listening! Even by YOUR OWN admission you are only applying certain bits of science, which ISN’T science, it’s pseudoscience. Furthermore, the science you don’t apply/account for, you dismiss on the basis that you don’t/can’t perceive the negative effects. This contradicts science because they do exist, we can measure them and science demonstrates they are audible. So STOP saying science supports your opinions because it does the opposite!

4a. So much flawed reasoning indeed! Your perception is what matters to you BUT it doesn’t matter to me or to science!! Excellent example of flawed reasoning, thanks for demonstrating!!

4c. Your reasoning contradicts science because although it agrees with certain parts of science, it dismisses and contradicts the demonstrated science that perception/preferences will vary from person to person! How many times???
4c1. Which is an OBVIOUS FALLACY because you are basing your reasoning on perception you admit you “can’t trust”, duh!
4c2. No, I think you’re “some kind of idiot” for basing your reasoning on only part of the facts, ignoring and contradicting others and then trying to ram your fallacies down everyone’s throat, in a science forum of all places!!!

The rest of you post is just more of the same ridiculous nonsense, self contradictions and ego massaging. Round and round you go, bla bla bla!

G

2. If I am using Newton's Law of Gravity to describe how an apple drops from a tree do you say I am wrong, because I ignore Einstein's General Relativity? Hopefully not, because the difference is meaningless. Newton's law is totally fine if you are using it where it is applicable. I am using the relevant parts of the human spatial hearing where they are applicable. When you put speakers in a room you are instantly ignoring a million small things, reflection from the coffee table for a start. People just don't care. Their ears get used to it. What crossfeed does is vanilla stuff compared to the "horrors" of speakers in a room. I know when/what to dismiss.

2a. I'm confident if Björk heard my choice for the proper crossfeed level she would approve it! Björk sounds very different on speakers and on headphones. Which one is her intent? If it is headphones without crossfeed, speakers are completely wrong (no excessive spatiality) and need crosstalk canceling!

2b. Applying science selectively doesn't make it speudoscience. How often can we apply science completely? 10 % of the time? Does that mean 90 % is pseudoscience? I haven't said I can't hear negative effects. I have said to me the negative effects are insignificant compared to the positive. To me correct ILD is much much much more important than correct ITD and science of spatial hearing tell us why. This is again knowing how to apply science.

4a. Science doesn't care about my personal preferences or how I perceive sounds, but it seems to nevertheless agree which is nice.

4c. You are scraping the bottom of the barrel. You sound desperate. Perception does vary from person to person, but there are common traits. You seem to think repeating this "you are dismissing things" argument is an effective tool against me. It isn't, because I can say I know how I dismiss things and how insignificant they are.

4c1. No, I am basing my reasoning on science which made me discover crossfeed in the first place. My perception, trustful or not agrees with the science. if that wasn't the case, I would have to trust science more than my perception.

4c2. Explain how I am contradicting other facts? I admit "ignore" of insignificant things, but not contradiction. Your posts are full of terms such as flawed, fallacy, contradiction etc. and a lot of talk about how science has to be used in it's entirely, but very little do I see science in action. Science means math. Where's your equations to demonstrate how "flawed" my thinking is? We can analyse each ignored part of the science one by one if you want. I am not afraid. I made the calculations 7 years ago and found out the things I ignore can be ignored. Intuitively it was pretty clear that's the case, but I wanted to be sure.

 

[71 dB]

ALL HEADPHONES HAVE INTEGRATED CROSSTALK CANCELING!

Yes, all headphones have crosstalk canceling and by default it is switched on!

Headphones don't have electrical crosstalk canceling circuits so it can be hard to notice it, but headphones manage to do crosstalk canceling by bringing the transducers very near the ears so that the amplitude difference in crosstalk is huge due to distance attenuation/head shadowing reasons. So, people listen to crosstalk canceled sound on headphones, most of them totally unware of it. Crosstalk canceling is beneficial only when the sound is binaural in nature. Music mixed for speakers doesn't need crosstalk canceling. Crosstalk canceling can mess up the spatiality creating totally unnatural levels of ILD at low frequencies for example. To turn headphone crosstalk canceling off and hear the music as intented by the artist, there are two things one can do:

1) Listen to speakers instead of headphones.
2) Use crossfeed or other method like HRTF convolution to reduce ILD. This "neutralizes" the crosstalk canceling.

 

[gregorio]

4a. Science doesn't care about my personal preferences or how I perceive sounds, but it seems to nevertheless agree which is nice.
4c. You are scraping the bottom of the barrel. You sound desperate. Perception does vary from person to person, but there are common traits.

4a. And here, yet again (!), you give a great demonstration of how your “reasoning” is fallacious! Yes, science does agree with your perception, NOT because your explanation is correct but because the science states that perceptions will vary. Science also agrees with what I’m perceiving, which is significantly different from your perception, for the same reason. The difference is that you’re contradicting this demonstrated science .....

4b. You seem to agree in principal with the science, that perception will vary from person to person, but then dismiss/contradict what you’ve just admitted by stating there are common traits. You’re effectively trying to assert that perception does not vary from person to person, that your perception is correct and different perceptions are wrong because you’ve thought about it a lot, don’t know anything about music production and therefore falsely apply the science of what occurs in nature! As previously, all you’ve demonstrated is that your insults actually apply to you! When are you going to stop doing this to yourself (and us)?

The rest of your post is just the same made-up personal opinion and self ego massaging as countless times before, yawn. You say you want things to change but then continue doing exactly the same thing, you think maybe we or science itself is going change if you contradict it enough times?

G

 

[bigshot]

Crossfeed is not an alternative to speakers any more than a rock is an alternative to mashed potatoes. They aren’t the same thing.