Analog crossfeed implementations

A song in osu! startled me the other day with some brutal hard-panning, and I'd prefer not to run DSP for a rhythm game that requires low audio latency. So it got me thinking about analog implementations of crossfeed, something I could put between a DAC and a headphone amp. Might also be useful in a studio, since latency seems like a big deal for anyone monitoring themself live, especially vocalists. This is a bit of a brain-dump and link-dump so I can get my mind elsewhere and also close all these tabs I've got open on my phone.

Benjamin Bauer: The OG, as far as I know. RLC circuit, therefore probably not something you're ever going to see in a portable device.

Siegfried Linkwitz: Like Bauer, designed for between an amp and high-impedance headphones

Jan Meier, "natural crossfeed": Not sure it's kosher to link to an archived page showing the design, considering Meier is still designing, building, and selling amps using this circuit

Jan Meier, "enhanced bass": Tweaked original design to mimic Linkwitz frequency response

Ingvar Ohman: I wasn't able to find the original article, so here's another page that used Ohman's circuit as part of a headphone amp. Seems like a lot of attention went to getting the time delay just right. Leaning toward building this one for my own use.

John Conover: That's a lot of op amps

Dimitri Danyuk: Potentiometer, neat!

DIY-Audio-Heaven: Nifty project but probably not the route I'm going to take

Pow Chu Moy's Modified Linkwitz: Tweaks things here and there but never claims to be a new design, also documented a little bit here

HeadRoom: Apparently an IC-based implementation of the filter Bauer was targeting? My old Total BitHead seems to have vanished into the aether so I snagged a cheap AirHead to crack open. I might want to reverse-engineer this circuit. And by "reverse-engineer" I mean blindly copy with no idea what any of it actually does, cargo cult coding style.

SPL Phonitor Matrix: No idea how this one works, but I've got their HPm on my wishlist for whenever I inevitably get into 500-series pro audio stuff. Seems like they do the processing on an unbalanced signal even in products with balanced headphone output.

Anybody know of more recent work? Tyll's data from when he took his rig to Harman's lab might be useful, for tuning a filter to a reference measurement head in a reference listening room. I still have the raw data that he made publicly available, and I swear he included data from hard-panned audio. I guess the ideal crossfeed filter would match the difference between the same-side ear and opposite-side ear, while mostly ignoring high-frequency wiggles due to external ear shape.

Room simulation with head tracking is a whole 'nother can of worms that I think is best considered out-of-scope for this thread.

 

I have the SPL Phonitor XE. I haven’t the knowledge or technical expertise to talk about how they achieve the effect, so I will touch on how they implement it and my opinions.
You have a variety of options on how wide you want the stage sound. I just have it set to max, but it’s not like you immediately feel like you are at a concert or anything like that. I’m one of those weirdos that actually like the ‘super stereo’ effect you get from headphones, so I really only use the crossfeed for classical music occasionally. You can definitely hear a difference - but nothing to make you gasp ha ha. I think I have said somewhere before, that I think the main benefit would probably be less fatigue over long listening sessions IMHO. Take it with a grain of salt as I don’t use it a lot. I can say unequivocally though that the Phonitor is an absolute beast.

 

Not something you can put between your DAC and amp but Decware’s implementation of crosssfeed on his amps is supposed to be one of the best and quite unique. There’s a loaner for a Decware amp you can sign up for if interested.

I don’t think he makes a separate cross feed unit but you could request one to express interest.

 

Would listening in mono be more acceptable than bad crossfeed? I do get annoyed sometimes by the hard channel separation with headphones and I wonder if I’m better off with mono.

Don’t know if the Schiit gadget is going to get released one day but I was hoping Mike would incorporate digital domain crossfeed along the frequency shifting he wanted to implement.

The frequency shift is kind of a hard sell however most people understand crossfeed and it seems like something that a company specializing in headphone amps would want to have.

 

I really like the crossfeed on my Phonitor. I used to have a Meier Aria that I thought was good, but that's so long ago I wouldn't count my memory as reliable on that.

I think Jriver's crossfeed is alright as it is fairly subtle. I don't remember the names of the other VST crossfeed plugins I've tried, but most did not really seem better than Jriver's to me. There was one that I did like years ago called Isone Pro by Toneboosters, but I don't think it's available anymore.

 

Unfortunately, it’s not an easy question. When you say “mono” I believe you intended a “counterfactual” product — in this case, a hypothetical recording if it is mono-mic’d. And it’s NOT obtainable from simple summing or averaging. This is because channels in stereo recordings are NEITHER identical nor mutually-independent. Sum of two signals would result in unwanted interference (over counterfactual mono recording) very likely. Some less annoying, others more so, etc. Compensating process is necessary to avoid this, which introduces quite a lot subjective decisions as inter-channel association is also uncertain. Professional grade signal processing for mono creation might consider this, but I’ve not seen this compensation in casual components that are used in many average player softwares. I’m guessing this applies to comsumer-target hardwares, too.

Long story short. It may work… or may not. That’s highly recording dependent — more accurately speaking, how stereo signals were mic’d, mix’d, and mastered would be crucial. Headphones base tonality and personal preference will be another HUGE deciding factors, too.. As an aside, I prefer xfeed to naively done monorization.

I do think what I said above is indeed applicable for all cross-feed discussion. Xfeed is eventually characterized as summing two channels with HPF/LPF (or some sort of shelving ones) and/or phase delays. How to set pass filters, how to add some more eqs, etc are all non-trivial questions.