[Hifihedgehog]

I am looking for information behind the math of crossfeed in order to better configure crossfeed filtering to the actual position of my headphone drivers. My thought process has been that crossfeed would very likely sound better, each spectrum (bass, midrange and treble) lining up and providing a more unison, dynamic sound if the delay, channel separation and shelf cutoff were optimized to the actual angling of my headphones. Here, this would mean that angling would be 150 to 180 degrees depending on the model in question. Part of my theory is, even at a wide angling situation as this, there is still not complete bass and midrange isolation between channels or sides like headphones produce. The real life scenario I am considering for this assumes music is commonly recorded with the microphone within three inches of the projection source (voice, instrument, etc.).
 
Now, as for your requested help, the methodology for me to figure out these numbers would involve a dummy head with two microphones mounted to the locations where the ears would be. As I have no access to a dummy head or its associated equipment to monitor the timing and leveling of the frequencies in such a configuration, I am wondering if there is a mathematic function I could work with to obtain some theoretical values and perhaps gauge what settings would be best for this. Thank you.

 

[castleofargh]

I'm really not a pro in this, but the way I see it what matters is the distance between your ears, not much where the drivers are. but maybe I didn't get what you're trying to do?

 

[Hifihedgehog]

I'm really not a pro in this, but the way I see it what matters is the distance between your ears, not much where the drivers are. but maybe I didn't get what you're trying to do?

Most crossfeed filters default to a setting which would simulate speakers being 60 degrees apart and I am trying to figure out is which settings would work to simulate a 150 to 180 degree speaker setup.

The delay you are referring is, to my knowledge, actually independent of speaker distance, since the delay concerns, for all intents and purposes, the delay between the bass and midrange, and the treble, and not the distance from the speakers to the listener.

That may sound weird but the way lower frequencies are projected, being higher wavelengths, they are long enough that they can wrap around your head, each at different wavelengths meaning they would reach the ear at different times. Higher frequencies remain more localized to just one side of your head because they have a shorter wavelength.

 

[castleofargh]

I never thought about crossfeed that way ^_^.
for me the delay is only to recreate the process of the left sound also reaching the right ear in live sound so only the size of your head matters. and the 60degree(to simulate speakers) is a result of how much left sound goes into the right and vice versa. not much a delay thing.
full mono being center in front of me, from there, attenuating the crossfeed value(left sound going right and vice versa) will "move" the origin of sounds slowly to the sides. and left and right totally separated(no crossfeed usual headphone sound) placing the source of the sound at 180°.
and then obviously the song has it's own positioning cues.
 
the matter of filtering some frequencies out of the crossfeed(at least removing the delays), to me was to avoid putting too much delays on the lead singer. because he will most probably already be in mono in front of me, and adding a delayed crossover will only make everything repeated twice like an echo. that's not something we want to do.
 
but again I've read something about it on meier's website once, and that's it. so although I came up with some reasons, I have zero certainty that they are right ^_^.
I'm sure someone will step in at some point. or you could try to contact xnor on another website(he's banned from here) he made a crossfeed component for foobar2000 some years back so he should know a good deal about it.
 
edit: here is what I saw, I will read it again with you maybe I'm telling stupid http://www.meier-audio.homepage.t-online.de/crossfeed.htm

 

[Hifihedgehog]

That's the general idea! There is a delay for bass and midrange frequencies reaching opposite channels at attenuated volumes. This in crossfeed helps to remove that awkward double subwoofer sensation (that is how I describe how I experience it) from hard panned sounds on the extreme left and right of the soundscape.

Now that I am at a proper computer and not confined to a smartphone, I can explain. I am very familiar with the basic theory behind crossfeed. I have read Jan Meier's articles on crossfeed, both on his site as well as the more detailed ones on Headwize. I just have not been able to find a clear cut equation that would allow me to adjust a digital crossfeed filter to a 150 to 180 degree speaker configuration as I have described above. There are a few functions in the old Headwize articles by Dr. Meier but they seem to deal more with the circuit designs than the delays and shelf cutoffs I am after. I would presume that the variable with the largest change would be the shelf cutoff since less lower frequencies would reach opposite ears if speakers were set up in a 150 to 180 degree configuration, but this is, of course, merely speculation. At any rate, I am after some clear cut equations for this and hopefully I can somehow come up with them.

 

[MindsMirror]

The delay should be a constant time delay for all frequencies. If you are calculating the theoretical delay for speakers at 180 deg, it should be just the time delay for sound to travel the distance between each ear. For example if your ears are 0.2m apart the delay (for all frequencies) at 180 degrees is 0.2/340 = 588us. For other angles you will have to do some trig/geometry calculations to work out the relative delay. I believe the only reason you need to handle certain frequencies differently is because of the amount of attenuation that occurs between your head. Certain frequencies may pass through or around your head with different intensity, but I believe they should all arrive with the same relative time delay, unless you want to account for sound that might curve around the head or take a different path, and I think that would be far too difficult to get right.
 
I don't think any of the attenuation would be to eliminate repeating the center channel audio. If you listen to mono audio on stereo speakers you do hear the delayed version of the left channel in the right ear, it sounds completely normal, and the crossfeed should emulate that just as well.

 

[Hifihedgehog]

  The delay should be a constant time delay for all frequencies. If you are calculating the theoretical delay for speakers at 180 deg, it should be just the time delay for sound to travel the distance between each ear. For example if your ears are 0.2m apart the delay (for all frequencies) at 180 degrees is 0.2/340 = 588us. For other angles you will have to do some trig/geometry calculations to work out the relative delay. I believe the only reason you need to handle certain frequencies differently is because of the amount of attenuation that occurs between your head. Certain frequencies may pass through or around your head with different intensity, but I believe they should all arrive with the same relative time delay, unless you want to account for sound that might curve around the head or take a different path, and I think that would be far too difficult to get right.
 
I don't think any of the attenuation would be to eliminate repeating the center channel audio. If you listen to mono audio on stereo speakers you do hear the delayed version of the left channel in the right ear, it sounds completely normal, and the crossfeed should emulate that just as well.

All this makes perfect sense. In other words, a 180 degree crossed-applied signal should have a constant delay since there is no curved path or arc to travel of varying delays like the illustration on Dr. Meier's site shows. I still have the matter of the shelf cutoff and attenuation to figure out. Thank you!
 
As for the delay I mentioned, the reason Dr. Meier uses a variable frequency delay is to reduce the comb effect.  It has been a while since I thoroughly read his synopsis but I caught the part where it speaks of dips in the higher frequencies in the mono channel when a constant delay is used. It makes some sense to do this since some delayed upper frequencies could cancel out their counterparts, resulting in a comb-like appearance to the effective frequency response. However, I would tend to think that this effect would be minimal if the upper frequencies in the crossfeed signal were attenuated enough.

 

[MindsMirror]

  As for the delay I mentioned, the reason Dr. Meier uses a variable frequency delay is to reduce the comb effect.  It has been a while since I thoroughly read his synopsis but I caught the part where it speaks of dips in the higher frequencies in the mono channel when a constant delay is used. It makes some sense to do this since some delayed upper frequencies could cancel out their counterparts, resulting in a comb-like appearance to the effective frequency response. However, I would tend to think that this effect would be minimal if the upper frequencies in the crossfeed signal were attenuated enough.

Wouldn't the comb effect also be present while listening to speakers? I think as long as you get the attenuation right, with a constant delay you should be able to emulate exactly what it would sound like listening to speakers. The varying delay might be used to produce a more neutral frequency response and more accurate monitoring, rather than a true emulation of listening to speakers. I guess it just depends what you want or what sounds best to you.

 

[Hifihedgehog]

  Wouldn't the comb effect also be present while listening to speakers? I think as long as you get the attenuation right, with a constant delay you should be able to emulate exactly what it would sound like listening to speakers. The varying delay might be used to produce a more neutral frequency response and more accurate monitoring, rather than a true emulation of listening to speakers. I guess it just depends what you want or what sounds best to you.

Again, that makes more sense. In a way, it is almost as if you just pulled a mythbuster on what is the basis of the Meier crossfeed. Counteracting the comb filter is exactly what Jan Meier is trying to do with his crossfeed as he describes here ( http://headwize.com/?page_id=654 ) in his article on Headwize, but choosing that as a design goal sounds wrong when you consider no speaker will ever be truly flat from a listener's position. I suppose any theory can sound good if you make the right (or wrong) assumptions.

 

[castleofargh]

   
I don't think any of the attenuation would be to eliminate repeating the center channel audio. If you listen to mono audio on stereo speakers you do hear the delayed version of the left channel in the right ear, it sounds completely normal, and the crossfeed should emulate that just as well.

lol I'm dumb. I never though about that. ^_^
I did well mentioning that I wasn't knowledgeable on the subject.

  I really am not.