• Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

How XCT Should be Done

I’ve been following the Smyth Realiser since they launched the Kickstarter. It’s a concept that feels a lot like an advanced version of XTC plus DCH.
It is neither.
You can only copy the acoustics of the physical situation at hand. So, if you want XTC you need a divider or a Processor as it is not possible to manipulate the signal flow in the Realiser afterwards. If the microphones do record crosstalk, it will be in the 'PRIR'.
If you want a “domestic concert hall“ you would have to record the ‘PRIR‘ in one.
And as the characteristics of your head and ears are part of the situation (and you want that reproduced) you more or less have to do all this personally. And if the speakers at hand have flaws, this will become part of the ‘PRIR‘, too.
From what I remember, they initially offered the two standard factory rooms (BBC and University of Surrey).
The BRIRs for these rooms are in the public domain, so yes they are available. However these are recorded with dummy heads and therefore one is listening with the “wrong ears“. And this does make a difference, for some more, for some less.
These rooms sound good, and together with the robust head tracking it is a nice experience.
The problem of the wrong head extends to all the available ‘PRIR‘, to overcome this, I would have to travel several hundred km and pay 1500$ to get a professional room measured with my head and it will become clear only afterwards how well this works out. How good are the speakers? It probably will not be bad, but is it worth 2000$?
If they can get impulse responses of great concert halls, they could even modify or personalise them further. Maybe I should email them with that suggestion.
As far as I know Smyth is struggling since the Kickstarter project as the price was not covering their costs at all. They probably don't have time and money to spent on extras for their legacy customers. As I understand it, their main market are professionals who want to work in their familiar acoustical environment on the go.
Their problem is that everything the Realiser was able to do long before its time is now also possible in a DAW or just a laptop for little investment, and actually one can do a lot more now. My Kickstarter-Realiser is mostly gathering dust nowadays.

I still don't get what the impulse response of a great hall is supposed to do for the reproduction of sound recorded in a great hall.
In my view, all the reflections and ambience of the concert hall, where the music was recorded, is already in the recording.
It is just not strong enough; if it were, it would not sound good in most listening rooms, as the listening room adds its own reflections. Therefore the microphones are placed directly at or on the stage and the sound is mixed accordingly. Recording with a stereo pair in row 8 would create a reverberant mess in most listening rooms.

The effect of using concert hall reverb would be like placing the (two!) speakers playing Beethoven (recorded in hall A) in a concert hall B. This is not what a concert hall is for and though it might work with some content, this will probably not sound like Beethoven played in that (which?) hall. Of course you can tune down the ambience to fit with the situation, but then, why use a concert hall impulse response in the first place?
What is needed in my view are superb listening rooms that add/amplify the ambient sound. If you have one available, you can record that situation with the Realiser. However, a room cannot actually amplify sound, so some kind of upmixing is necessary. [The Realiser could record that upmixing too, but now it gets complicated, because you might want to have more or less upmix, and this a new acoustical situation and you have to record again.]

All this can be done in a virtual room binauralisation with much more flexibility.
Beethoven on the beach (almost) like in the great hall it was recorded in? No problem, just a bit weird ;-)
 
  • Like
Reactions: STC
It is neither.
You can only copy the acoustics of the physical situation at hand. So, if you want XTC you need a divider or a Processor as it is not possible to manipulate the signal flow in the Realiser afterwards. If the microphones do record crosstalk, it will be in the 'PRIR'.
If you want a “domestic concert hall“ you would have to record the ‘PRIR‘ in one.
And as the characteristics of your head and ears are part of the situation (and you want that reproduced) you more or less have to do all this personally. And if the speakers at hand have flaws, this will become part of the ‘PRIR‘, too.

The BRIRs for these rooms are in the public domain, so yes they are available. However these are recorded with dummy heads and therefore one is listening with the “wrong ears“. And this does make a difference, for some more, for some less.
These rooms sound good, and together with the robust head tracking it is a nice experience.
The problem of the wrong head extends to all the available ‘PRIR‘, to overcome this, I would have to travel several hundred km and pay 1500$ to get a professional room measured with my head and it will become clear only afterwards how well this works out. How good are the speakers? It probably will not be bad, but is it worth 2000$?

As far as I know Smyth is struggling since the Kickstarter project as the price was not covering their costs at all. They probably don't have time and money to spent on extras for their legacy customers. As I understand it, their main market are professionals who want to work in their familiar acoustical environment on the go.
Their problem is that everything the Realiser was able to do long before its time is now also possible in a DAW or just a laptop for little investment, and actually one can do a lot more now. My Kickstarter-Realiser is mostly gathering dust nowadays.

I still don't get what the impulse response of a great hall is supposed to do for the reproduction of sound recorded in a great hall.
In my view, all the reflections and ambience of the concert hall, where the music was recorded, is already in the recording.
It is just not strong enough; if it were, it would not sound good in most listening rooms, as the listening room adds its own reflections. Therefore the microphones are placed directly at or on the stage and the sound is mixed accordingly. Recording with a stereo pair in row 8 would create a reverberant mess in most listening rooms.

The effect of using concert hall reverb would be like placing the (two!) speakers playing Beethoven (recorded in hall A) in a concert hall B. This is not what a concert hall is for and though it might work with some content, this will probably not sound like Beethoven played in that (which?) hall. Of course you can tune down the ambience to fit with the situation, but then, why use a concert hall impulse response in the first place?
What is needed in my view are superb listening rooms that add/amplify the ambient sound. If you have one available, you can record that situation with the Realiser. However, a room cannot actually amplify sound, so some kind of upmixing is necessary. [The Realiser could record that upmixing too, but now it gets complicated, because you might want to have more or less upmix, and this a new acoustical situation and you have to record again.]

All this can be done in a virtual room binauralisation with much more flexibility.
Beethoven on the beach (almost) like in the great hall it was recorded in? No problem, just a bit weird ;-)

I revisited the Head-Fi threads where many posts about the Smyth Realiser can be found. There may be some confusion.

There is no acoustic crosstalk with headphones. If anything, people sometimes add a little crossfeed because pure isolation can feel unnatural — just as high attenuation is not necessarily better with loudspeaker XTC.

The patent clearly describes how the system is implemented. What is less clear to me is whether the ear measurement always uses the user’s real pinna response, or whether it can also involve a more generic correction matched to the measured pinna in a manner similar to Apple Spatial Audio.

Regarding the PRIRs and BRIRs I use: none of them were made by me. They were originally created for commercial or research purposes. Most of the information I have shared comes from the Realiser manual, the patent, and user reports (including material that used to be on their YouTube channel — not sure if it is still active).

Smyth was once used as a reference during the development of a crosstalk-cancellation plugin by a company where I played a small role testing the software. That is the extent of my direct involvement.

Coincidentally, I believe Timothy posted something years ago about asking whether crosstalk cancellation could be incorporated. I am not sure of the exact context or how the developer responded, but it would be interesting to know more. Topics like interaural crosstalk, natural crosstalk, and how systems such as the Realiser and Apple Spatial Audio actually work are still areas where clarification is useful.

That’s all I can usefully add at this point.

ST
 
For listening tests, I generated a simple signal consisting of a sequence of pink noise bursts (L→C→R→C, looped). The objective was to find the settings which result in the least timbral difference between the phantom center and the hard-panned sounds (which are unaltered by the XTC). At least in my particular setup and for the particular test signal used, the optimal ILD seems to be approximately 3-4dB higher than that suggested by the anechoic data, which is similar to what @RaulFO found in his tests.
Regarding the point for which I was quoted, I see something similar in my own experiments: in a normal domestic room, the attenuation that gives useful XTC results tends to be somewhat greater than what simple anechoic models might suggest.

My tentative explanation is simply that the room contributes a considerable amount of delayed acoustic energy. This makes perfect cancellation of the direct crosstalk path impossible, but it may also make some of the artifacts associated with strong cancellation less perceptually dominant. I would regard this only as a working hypothesis, not as a firm conclusion.

There is another aspect of the discussion that I am finding increasingly difficult to follow, however.

I think it is important to distinguish between describing an acoustic situation and inverting an acoustic transfer function.

Suppose I measure the complete 2x2 BRIR matrix of my loudspeakers, room, head and ears. If I design an approximate inverse of that matrix and use it to drive those loudspeakers, I have essentially designed a transaural system: room/speaker equalisation and crosstalk cancellation can both be part of that inverse.

That is one problem.

If instead I convolve signals with the original BRIR and reproduce them over headphones, I am doing something quite different: I am trying to reproduce at my ears the acoustic situation that was measured.

That is binaural reproduction.

Now imagine applying both operations in succession: first the inverse BRIR and then the original BRIR. In the ideal mathematical case,

H^{-1} * H approx I

so I have first removed the acoustic system and then recreated it — or vice versa.

At that point I think we should ask what problem we are actually trying to solve.

XTC is not itself an acoustic environment that has to be “reproduced”. It is a compensation required by loudspeaker reproduction because each loudspeaker reaches both ears. With headphones that acoustic crosstalk path is essentially absent.

Of course binaural rendering and XTC can be combined in a perfectly meaningful way:

binaural renderer → desired left/right ear signals → XTC → loudspeakers.

In that architecture the binaural stage defines what should arrive at the ears, while the XTC stage solves the loudspeaker delivery problem.

What I find confusing is when BRIR reproduction, headphone crossfeed, loudspeaker crosstalk cancellation and binaural rendering start being discussed as if they were different versions of the same operation. They are related, but they solve different problems.
 
My tentative explanation is simply that the room contributes a considerable amount of delayed acoustic energy.
My hypothesis is along the same lines. Correcting the direct spectrum inevitably results in a "peaky" indirect spectrum, which is presumably audible as coloration in a normally-reflective room when something approaching full correction is attempted. It would be interesting to repeat the experiment in an anechoic or near-anechoic space, but I don't have easy access to such a space.
 
What I find confusing is when BRIR reproduction, headphone crossfeed, loudspeaker crosstalk cancellation and binaural rendering start being discussed as if they were different versions of the same operation. They are related, but they solve different problems.
They are all inter-related and so end up being discussed together. What are we trying to do with headphone crossfeed? A big part of it is reproducing crosstalk interference and BRIR. So should I try to make headphones sound like a standard stereo pair in a typical good room, or should I take advantage of the lack of crosstalk and only selectively apply the crossfeed?

One thing I've been taught at my place of employment is that crosstalk from reflections is also damaging to stereo perception. So this leads to the idea of a BRIR that only has reflections from each speaker in the correct ear - more precisely, only arriving at the correct ear first and crossing the head from that direction. We actually want both ears to hear the reflection, I think. This would be a sort of magical room that doesn't allow any cross reflections to reach the wrong ear first. I've always wondered how important that really is. Some experiments I've done setting up a room with absorbers and deflectors angled to do this as much as possible did seem to widen and strengthen the stereo image despite not addressing direct crosstalk from the speakers straight to the ears.

I guess I need to learn how to set up the SPARTA software or some other software to simulate such an environment and see what it produces over headphones. If just controlling the reflections really works, then maybe that's a better approach than actively manipulating the signal to cancel crosstalk at each ear. If it works in a real room then it we should hear it over headphones too using BRIR that mimics the situation. It does make a lot of sense that in a symmetrical listening arrangement, the reflections from each side of the room will arrive at the corresponding ear at the same time, so those reflections should reinforce the imaging, especially if there are no cross reflections adding to the confusion by crossing the head from the wrong direction.

One thing I realize about using a center divider to do crosstalk reduction in a real room is that it does absorb a lot of the cross room reflections as well as absorbing much of the direct crosstalk. This may be why the effect was strongest when the divider plate was large and went all they way between the speakers and behind them. Otherwise a simple, small local barrier near the head should produce similar results, but it doesn't.
 
Last edited:
Of course binaural rendering and XTC can be combined in a perfectly meaningful way:

binaural renderer → desired left/right ear signals → XTC → loudspeakers.

This statement is confusing.

Binaural rendering is the process of taking a non-binaural recording and turning it into a signal that sounds like a binaural recording when played over headphones.

XTC, by contrast, is what allows a binaural recording (which is still a stereo signal, but one that already contains more complete interaural and spectral cues) to be reproduced over loudspeakers while preserving those binaural characteristics. In doing so, XTC can also make ordinary non-binaural stereo sound more binaural-like, using only crosstalk cancellation. Different problems and different solutions for making the sound to sound binaural.

ST
 
I would frame it as a clarification rather than a correction, because I think part of the disagreement comes from using the same vocabulary for different stages of the reproduction chain.

I think the point where we differ is mostly conceptual rather than mathematical.

Binaural reproduction and XTC certainly have a common origin: both ultimately deal with the transfer from sound sources to the two ears, and both can be described using HRTFs, HRIRs, BRIRs or 2x2 acoustic transfer matrices.

But sharing the same underlying acoustic model does not make them the same technique.

A binaural renderer tries to answer:

What signals should arrive at the left and right ears in order to represent a particular acoustic scene or source direction?

XTC tries to answer a different question:

Given two loudspeakers which both reach both ears, what signals should I send to those loudspeakers in order to reduce the unwanted cross-paths?

Those two operations can obviously be placed in the same chain:

virtual/binaural scene → desired ear signals → XTC/transaural stage → loudspeakers

In that sense they are complementary.

But I would not say that XTC turns ordinary stereo into binaural.

XTC can make conventional stereo produce a much wider soundstage, reveal interaural differences that loudspeaker crosstalk was degrading, and generate virtual positions well outside the loudspeaker span. But it does not create binaural cues that were never present in the original recording.

A simple example is front/back localization.

Imagine a true binaural recording made with a dummy head facing away from a sound source. That two-channel recording contains the spectral and interaural cues associated with sound arriving from behind the head.

Over headphones, those two ear signals can be presented directly, so rear localization can be reproduced over a full 360° perceptual space, at least to the extent allowed by the quality and individual match of the binaural recording.

With a conventional pair of loudspeakers in front of the listener, the situation is quite different.

XTC can cancel or reduce the direct cross-paths between the loudspeakers and the opposite ears, but the physical sources are still located in front of the listener. The acoustic field generated by those loudspeakers is therefore not equivalent to having an actual source behind the listener.

In a very idealized two-point model, one can of course write a 2x2 transfer matrix and attempt to reproduce arbitrary pressure signals at the two ears. Mathematically, that is perfectly legitimate.

But that is not the same as saying that two frontal loudspeakers can reproduce an arbitrary binaural 360° scene in a real room with the same perceptual robustness as headphones.

Front/back discrimination is not determined only by ITD and ILD. Spectral cues from the pinnae, head and torso are important, especially at higher frequencies, and those cues are highly sensitive to head position and orientation.

And there is another practical problem: the room itself.

The loudspeakers are physically in front of the listener, so the early reflections and the diffuse/reverberant field are generated by frontal sources. Even if the direct field at one head position has been manipulated to approximate some binaural target, the reflected field still carries information about the actual physical source geometry.

In other words, the room can provide spatial evidence that contradicts the virtual direction one is trying to synthesize.

This is one of the reasons why I think it is useful to keep binaural rendering and XTC conceptually separate.

Binaural rendering is primarily about creating spatial information.

XTC is primarily about delivering two signals through loudspeakers while reducing the contamination between them.

They may use the same HRTF/HRIR mathematics, but after that common starting point their objectives are different.

So when you write that XTC can make ordinary stereo sound “more binaural-like”, I understand what you mean perceptually: reduced crosstalk can produce a much stronger sense of width and spatial separation.

But technically I would still distinguish the two things.

A wider and more spatial stereo image is not the same as a binaural scene.

For me that distinction is important because otherwise BRIR reproduction, headphone crossfeed, binaural rendering and loudspeaker crosstalk cancellation start to become different names for different parts of one very large problem, and then it becomes difficult to know which transfer function we are modelling, which one we are inverting, and what result we are actually trying to obtain.
 
There is a practical experiment that may help clarify why I prefer to keep “binaural” and “XTC” as separate concepts.

In the NatAmbio testing_XTC tools I use two different kinds of synthetic test signals. One is constructed from binaural/interaural cues, using the usual ILD/ITD relationship associated with source direction. The other is deliberately not binaural: its spatial behaviour is generated only from conventional stereo variables such as level panning and correlation/anticorrelation.

The interesting point is that, when reproduced through an XTC loudspeaker system, both signals can produce a very wide and clearly structured spatial scene. In the frontal horizontal region they can even reach similar perceived positions by completely different routes.

That is precisely why I would not use “binaural-like” as a technical description of what XTC is doing.

A signal based on panning plus correlation/anticorrelation contains no measured or synthesized HRTF, no pinna filtering and no explicit binaural encoding. Nevertheless, XTC can map those L/R relationships into large apparent source displacements, including positions well outside the loudspeaker base.

If the same signal is listened to through headphones, however, the distinction becomes obvious. Level panning mainly moves the image along the interaural/intracranial axis, while strong decorrelation or anticorrelation can make it diffuse or difficult to localize. It does not suddenly become a properly externalized binaural source at the corresponding azimuth.

A genuinely binaural signal is different because the directional information is already encoded in the two ear signals. Headphones can reproduce those signals directly. XTC may also be used as a transaural delivery mechanism for some binaural material, but that does not make XTC itself a binaural renderer.

So, at least for me, there are three different concepts here:

binaural encoding — how directional information is represented in the two ear signals;

stereo spatial encoding — level, phase, correlation and anticorrelation relationships between two channels;

XTC — how those two channels are reproduced through loudspeakers while reducing the unwanted cross paths.

Their perceptual results can overlap, especially for frontal localization, but that does not make the underlying signals equivalent.

In fact, this overlap is one of the things I find most interesting about XTC: a completely non-binaural synthetic stereo signal can produce very high spatiality when reproduced with crosstalk cancellation.

This is also why I would distinguish “spatiality” from “binaurality”. A signal can be highly spatial without being binaural, and removing loudspeaker crosstalk does not create binaural information that was absent from the source.

I have included the test generators here, for anyone interested in experimenting with the two approaches:

NatAmbio — tools/testing_XTC
 
One more simple example may help illustrate the distinction.

Think of the well-known binaural recordings where someone moves a pair of scissors around a dummy head. Over headphones, the scissors can appear to be only a few centimetres from your head, almost as if someone were actually cutting your hair.

If you play that same two-channel recording through a pair of frontal loudspeakers, even with XTC, that near-field distance percept is not automatically reproduced.

The reason is simple: the binaural recording contains not only left/right differences, but also distance-dependent near-field cues, pinna/head filtering, very strong changes with small movements, and a particular relationship between direct and reflected sound.

With loudspeakers, the physical sources are still several metres away and in front of the listener. The room reflections correspond to those real source positions, and small head movements do not produce the same geometric changes that a real source a few centimetres from the head would produce.

So again, XTC may reproduce a very wide and convincing spatial image, but high spatiality is not the same thing as binaural reproduction.

A binaural recording can encode azimuth, elevation and distance cues. XTC, by itself, does not create those cues if they are not already present, nor does it guarantee that all of them can be reproduced perceptually through a conventional frontal loudspeaker setup.

That is another reason why I prefer to keep the two concepts separate: binaural describes how the spatial information is encoded; XTC describes how two channels are delivered through loudspeakers.
 
My personal experience is with the RACE processor as implemented in the most current version of camilladsp. It's part of a piCorePlayer installation on an RPI4. The RPI4 is configured as a player in an LMS system, feeds an RME ADI-2 DAC that in turn feeds a Benchmark AHB-2 amp via XLR interconnects. I have a typical living room setup with two speakers (Ascilab C6B) and two SVS well-optimized subs. The speakers are about 10 feet from my listening position, a bit less than that apart, and speakers are toed in at 23 degrees.

As a complete non-technical person, after I read about the XTC concept on this forum, I undertook to give XTC a try just as a fun project to see what would happen. It took me about 3 weeks, probably 50 hours, total, a lot of false starts, and a lot of help from Gemini and CoPilot (who did not always agree!) and, at times, it was not so fun ;). But I eventually got there.

After getting camilladsp and the RACE processor correctly configured, there was a little bit of fiddling with EQ and adjusting the RACE processor settings to get the sound quality that was both distortion-free and of the tonality that I personally feel is right.

My results are just as RaulFO explained above. The soundstage has a larger, deeper spatial image, occasionally jumping outside the speakers (recording dependent), and noticeably increased clarity of instrument/voices and spacing in relationship to each other.

Overall, I really enjoy the effect and improvement in listening quality. I find that, other than left/right differences and reduced spatial cues, listing far off axis from the speakers still sounds pretty good---no echoing or other odd effects. At this point, after a week or so of listening, I don't feel there is anything negative about my use of XTC as I've implemented it. The only "problem" is that I'm now hearing defects on some recordings that have been masked up until now.
 
Last edited:
One more simple example may help illustrate the distinction.

Think of the well-known binaural recordings where someone moves a pair of scissors around a dummy head. Over headphones, the scissors can appear to be only a few centimetres from your head, almost as if someone were actually cutting your hair.

If you play that same two-channel recording through a pair of frontal loudspeakers, even with XTC, that near-field distance percept is not automatically reproduced.

The reason is simple: the binaural recording contains not only left/right differences, but also distance-dependent near-field cues, pinna/head filtering, very strong changes with small movements, and a particular relationship between direct and reflected sound.

With loudspeakers, the physical sources are still several metres away and in front of the listener. The room reflections correspond to those real source positions, and small head movements do not produce the same geometric changes that a real source a few centimetres from the head would produce.

So again, XTC may reproduce a very wide and convincing spatial image, but high spatiality is not the same thing as binaural reproduction.

A binaural recording can encode azimuth, elevation and distance cues. XTC, by itself, does not create those cues if they are not already present, nor does it guarantee that all of them can be reproduced perceptually through a conventional frontal loudspeaker setup.

That is another reason why I prefer to keep the two concepts separate: binaural describes how the spatial information is encoded; XTC describes how two channels are delivered through loudspeakers.

IMO/IME, a lot of it comes down to the perceived loudness, however, spatial quality as far as room is concerned, does require the right kind of manipulation of indirect sound.


The two examples, the same room, the same MLP (3,5 meter distance), both reproduce very clean direct path, close to head localization, enhanced depth perception and low frequency envelopment, but the first (the one where reflections are largely inaudible) does not reproduce spatial quality nor apparent source width (outside degrees of conventional stereo, not the actual listening triangle). It may also at times sound weird due to important cues that are missing, even lacking externalization.

The second example compensates for the missing cues but this is very room dependent, as far as envelope smoothing, lack of temporal masking and late arrival actually sounding right. So this example is no more than anecdotal, in case the same system being set up in any other room.
 
Is this simply a preamp with a mid/side matrix? These days it's very easy to get this functionality in vst plugins, and I guess you could implement your 'quad' thing if you had a DAC with 4+ outputs, although if I'm reading this right you're putting Mid/Mid on the front and Side/Side on the back?
I play with them because I have had them since they were new. The configuration stated is simply a factory suggestion. That does not mean that one must use them that way.
I use them many ways. with and without various processors, quad decoders, whatever (it varies how I am using them on a regular [ir-regular, perhaps] basis):
Apt Holman Preamplifier Review back panel Phono  Vintage Stereo.jpg

With 2 of these, there are many ways to configure them, using processor loops, external crossovers (usually for my subs), so, and many other things. My system is not static in it's configuration, it varies with what I feel like doing. It may stay a certain way for months, or may change daily. There is no computer or monitor connected to my system (unless it is inside another component, [and the only screens in my home are connected to my deck top {and for use with my region free oPPo 205 UDP, whose audio out goes to a Bluetooth transmitter, to a Bluetooth receiver connected to the stereo]}).
VST plug-in's are not even on my radar (maybe, one day but unlikely). As a computer within my stereo system is highly unlikely (at his time, anyway)
In 2007, my wife asked me "Do we really need a TV?" I said "not really." She got rid of it & we haven't had on since. There is a clue there about how I feel about computerized things in my home. Yes, for somethings they are necessary (including my one day wanting to have all my audio, from all my formats, digitized, on a home server, connected to my stereo [but not connected to anything on the outside of my home]).
For me, quite simply, I'm just having fun. (And this is good enough for me!):
Apt Holman Preamplifier Measurements Vintage Stereo.png



Wow, this is excellent! I expected SINAD in 60s or 70s and here we have 90 dB. It is dominated by second harmonic at that level. Even more impressive is the signal to noise ratio with same unity gain:
Apt Holman Preamplifier Measurements SNR Vintage Stereo.png



Frequency response: dead flat in audible band:
Apt Holman Preamplifier Measurements Frequency Response  Vintage Stereo.png


Crosstalk:
Apt Holman Preamplifier Measurements Crosstalk  Vintage Stereo.png


There is essentially no frequency dependency in distortion+noise:

Apt Holman Preamplifier Measurements THD+N vs Frequency Vintage Stereo.png
 
Maybe dumb question maybe not, if crosstalk is so objectionable why not listen on headphones?

Crosstalk coloration is accounted for in the mix because all stereo setups have some. So the comb filtering you get in the upper mids and treble should be handled already, at least in terms of tonality.



This sounds like my typical headphone listening experience... Which is why I use cross feed to avoid it ;).
I rarely listen to headphones: but (IMHO) cross feed is definitely helpful when doing so.
 
Back
Top Bottom