Can channel crosstalk be actually a good thing?

[Asylum Seeker]

Personally, I do not use cross-feed and have no use for it on my system since I prefer inaudible cross-talk on my sound reproduction system. Cross-feed defeats that purpose and thus present imaging that is not true to the source. IEMs and headphones have different perceived imaging so even cross-feed switches or DSP cross-feed will NOT correct the imaging of headphones to the way the sound is presented through speakers. The common theory that I know of is that cross-feed helps older recordings sound tolerable on headphones. I certainly enjoy the way my CA Andromeda IEM images the sound without any additional cross-feed whatsoever. BTW, I've demoed my CA Andromeda through Chord Hugo 2 with those cross-feed switches and yep they turn my CA Andromeda to crappy imaging (to my subjective opinion) after flipping the cross-feed switches or balls in this case to any of "low, medium, high" settings

View attachment 54660

Ok, we can safely assume that you don't know the theory of how crossfeed operates since, after being asked a few times, you are clearly unable to articulate it. Under discussion are not your personal preferences, your personal habits, your subjective opinion or specific equipment. Lest it still be unclear to you, this is an objective discussion about how the theory of cross-feed extends--or does not extend--to crosstalk in headphones. That you use or do not use, that you like or do not like, that you have tried or have not tried crossfeed has no relevance to the discussion.

 

[majingotan]

Ok, we can safely assume that you don't know the theory of how crossfeed operates since, after being asked a few times, you are clearly unable to articulate it. Under discussion are not your personal preferences, your personal habits, your subjective opinion or specific equipment. Lest it still be unclear to you, this is an objective discussion about how the theory of cross-feed extends--or does not extend--to crosstalk.

Lest it be turned to subjective and objective debate, I participate to learn as before I came to ASR, I was a naive hobbyist who didn't know a thing or two about measurements and other technical notes regarding audio reproduction. I get tidbits of valuable info from very technical participants in this forum and try to piece them with my limited knowledge. I do fail or get the grasp of some technical articles but unfortunately this one is a grey area for me as I've never really delve in to cross-feed and cross-talk (I've only become aware of this cross-talk definition after seeing measurements here for the 1st time) and my experience with the current cross-feed technology is rather unsatisfactory as it fails, in my experience, to provide speaker-like presentation (referencing the near field speaker listening) to my IEMs. I bolded that part as this is how my feeble brain interprets cross-feed where as Wikipedia states that "Directional sound perception is based on the delay between the same sound reaching a person's left and right ears. In stereo speakers, the sound from one speaker reaches both ears, although at different levels, and with a delay between one ear and another, since the speaker is placed away from the center. In headphones, this crossfeed does not occur, so the resulting stereo image is different from what is heard from speakers. A crossfeed signal processor attempts to recreate the stereo image heard from speakers by mixing some signal from the left channel into the right channel, and vice versa. The intent to produce speaker-like sound in headphones distinguishes crossfeed from the more general concept of stereo width reduction, which involves similar techniques." Maybe that's the statement that supports the theory that you're looking for.

Going back to original post then:

Theory - stereo imaging suffers in headphones because, unlike speakers, each ear is isolated from what the other ear hears.

Question - can some channel crosstalk--and I am not talking about 'crossfeed'--be actually a good thing for imaging in headphones/earphones?

If there exist say a crosstalk DSP setting for every headphone or IEM in the market that will recreate the speaker-like imaging perfectly then I'm all for it.

 

[Tks]

Theory - stereo imaging suffers in headphones because, unlike speakers, each ear is isolated from what the other ear hears.

Question - can some channel crosstalk--and I am not talking about 'crossfeed'--be actually a good thing for imaging in headphones/earphones?

Sure if you can determine what is actually leaking. I'm not really knowledgeable on what is occurring in every instance of crosstalk (if any distortions or things of that nature crop up as a result). So I can't really speak much to crosstalk desirability (since you're asking a subjective question in essence). Objectively, this is undesirable as it's not a toggle like crossfeed, thus an intrinsic issue just like even-order harmonics are (nontheless preferable to folks like tube aficionados). The reason for that is simple, is it an impediment to the aspiration of transmission devices (the most desirable aspect being the preservation of fidelity). So when you ask is it a good thing for some IEM's or Headphones, sure if you get lucky and get the amount of crosstalk you prefer or something. But otherwise, because it pollutes everything - it makes about as much sense as buying amps for exclusive use for each listening device you may have. It's basically a pragmatic issue.

Crossfeed I think in the digital domain is delaying signals on top of merging portions of each channel to one another. In the analogue, I haven't the faintest of clues of what is going on aside from just nullifying portions of the stereo recording itself I suppose.

Imagine having crosstalk that isn't to your flavor, and you try to run DSP Crossfeed on top of it.. That probably would sound pretty poor. This is why crosstalk can never be "good" aside from accidentally based on someones preference on the off-chance.

I hope that was of some use as a perspective with respect to your question.

 

[solderdude]

crossfeed and what I wrote about it

crosstalk usually also is somewhat frequency dependent but in general becomes more 'mono-ish' at higher frequencies where with crossfeed the lower frequencies are 'mono-fied' and higher frequencies are not.

The effect differs. Some swear by crossfeed. I only found it to be beneficial for some recordings. In some cases it 'removed' listening fatigue.
Now I do not use it anymore.

Then there is negative crosstalk which happens when low impedance headphones are connected via a relatively high resistance 3-wire cable or an amplifier with an unusually high resistance 'ground/common' connection in the amp itself.
This 'mono-fies' the stereo signals and subdues (cancels) the mono signals resulting in a weird stereo image.
This can easily not be seen when testing amplifiers because it is a cable/headphone effect which I never see 'tested' or measured.

Then there is crosstalk by mechanical origin. Needle angle opposite the groove and playback heads not being height adjusted.

 

[tuga]

Theory - stereo imaging suffers in headphones because, unlike speakers, each ear is isolated from what the other ear hears.

Question - can some channel crosstalk--and I am not talking about 'crossfeed'--be actually a good thing for imaging in headphones/earphones?

It's great if you are listening to the Beatles.

 

[solderdude]

It's great if you are listening to the Beatles.

unless they are mono recordings

 

[maverickronin]

I see crossfeed as only applicable to old recordings with extreme channel separation since the recording techniques back then are still simplistic. Modern recordings are much more complex in their mixes and as such modern DACs have evolved in a way to reproduce the mix as close to the source as possible. Sure you can toggle crossfeed on, but you aren't hearing the original source, but rather imaging that suits your subjective preferences. Then again, to me it's an equivalent of a tube effect box to me which is I'm also guilty of using.

Crossfeed is pretty much required for stereo recordings to sound anything approaching natural on headphones as stereo mixes for speakers assume crosstalk that increases as frequency deceases.

current cross-feed technology is rather unsatisfactory as it fails, in my experience, to provide speaker-like presentation

But this is also true as most crossfeed implementations (even most digital ones, which just simulate simple analog circuits) are more or less just a low pass filter on one channel fed to the opposite. They won't make headphones sound like speakers, but they do get rid of the "speakers set up at either end of a long hallway" effect that most headphones suffer from. If you don't get headaches from that, then more power to you!

More advanced HRTF simulations do a much better job at simulating speakers. TB Isone is an excellent example. It's designed purpose is to let you test you mix in different environments and on different kinds of speakers, but for straight music listening it's best if you disable the room sim, set the virtual speakers to flat, and then adjust the HRTF to match your own.

 

[sam]

I get your point, but I would never say uncontrollable cross talk is a good thing.
You could probably find one or two headphones that might sound good with the crosstalk, but it will probably be bad for rest of the others.
If you can mix the left and right signals with a specific gain, frequencies, and harmonics dynamically, then it's a different story.

 

[STC]

Of course you have complicating factors like the Ambiophonics guys - who spend a lot of time and money trying to REMOVE crosstalk from speaker systems. It can work spectacularly well on some recordings, but isn't really a general purpose technique.

It wrong to say Ambiophonics requires you to spend money. All you need is a DAW and you can make in yourself or just buy the $10 plugin.

It works fine with all stereo recordings.

 

[Deleted member 16543]

I think (intentional) cross-feed can be made good or bad, depending on how it is implemented. If done taking into account the head shadow effect and time delay of actual speaker reproduction cross-talk, it can help in making headphones sound more like loudspeakers.
If intended in the sense of a little of left channel bleeding straight through on the right channel and vice-versa, without any other consideration, then its results can be erratic, depending on the recording in question.
I am actually more interested in the opposite process, crosstalk removal from speakers' playback, to better enjoy binaural recordings over loudspeakers.
They are still quite enjoyable to me even without crosstalk cancellation, but could maybe be one of those cherry on top situations.

I think a major reason why stereo representation is somewhat limited to the width of L-R speaker placement is cross-talk itself.
On the other hand, a major reason for in-head sounds with headphones playback comes from the brain trying to make sense of the phase information it gets from L and R transient signals.
I can't find it anymore, but I think it was sr3d that had a demo video of some kind, that used a progressively sped up transient sound recorded binaurally. What happened there, as the recording speeds up more and more, was that the frequency content shifted upwards, but the initial phase difference between L and R remained unaltered. Up to a certain frequency, the initial phase difference correlates to a real possible location in space outside the head, which is where the brain puts the sound source as the recording speeds up more and more. Higher frequency = lower wavelength, so same phase for lower length means lower source distance. The effect is that of the sound source progressively getting closer to your head.
Until the frequency is high enough that the phase information doesn't correlate anymore with any point outside the head, so the brain puts the sound inside of it. And that's exactly what one hears in the video.
I'm not sure I'm explaining this clearly enough. Maybe somebody knows what video I'm talking about..
Anyway, my point being that with phase incoherent recordings, made out of multiple mono mics and mixed down to stereo, there might be a benefit in employing some well thought out cross-feed algorithm to simulate speaker's playback and get the soundstage out of the head. But it does depend on the signals being fed to L and R channels, and results may vary. Worst case scenario being with binaural recordings, where the correct phase is embedded in the recording to begin with, and cross-feed might actually be counter-productive.

 

[DonH56]

Assuming the same amplitude in L/R channels:

• Crosstalk in phase will increase focus on the middle (center) and reduce the image (worst case = mono).
• Crosstalk out of phase will reduce the center content and widen the image (worst case = "hole in the middle" effect).
• Random amplitude and phase will do something in between.

Various folk have implemented cross-fade schemes to try all of these out, natch.

 

[STC]

think a major reason why stereo representation is somewhat limited to the width of L-R speaker placement is cross-talk itself.
On the other hand, a major reason for in-head sounds with headphones playback comes from the brain trying to make sense of the phase information it gets from L and R transient signals.

Interaural Crosstalk is hearing the sound from one speaker in the opposite ear. Depending on the listening angle of speakers replacement in stereo production over loudspeakers will always produce a delay error of 220 μs ( at 60 degrees) . The same recording with headphones will not have the crosstalk ( interaural crosstalk NOT to be confused with channel leaking or crosstalk) error. Now we receive the correct ITD ( encoded in the recording) which will not be possible in speakers playback because the ITD cannot exceed the delay produced by speakers angle.

However with headphones you lose the role of pinna in direction finding which will cause the sound to be internalized. A sound is 3D when the ITD, ILD and pinna function are delivered in tact to the brain. A binaural recording sound 3D because it is recorded and reproduce at the same spots.

 

[Deleted member 16543]

Interaural Crosstalk is hearing the sound from one speaker in the opposite ear. Depending on the listening angle of speakers replacement in stereo production over loudspeakers will always produce a delay error of 220 μs ( at 60 degrees) .

Yes. There's also the head shadow effect to take into account, but ITD wise you are correct.

The same recording with headphones will not have the crosstalk ( interaural crosstalk NOT to be confused with channel leaking or crosstalk) error.

Correct. I believe the channel leak (intentional or not) was referred to as cross-feed by somebody else. My point is that if it is an instantaneous cross-feed, without the ITD 'error'/head shadow EQ, it's not going to possibly sound like speakers over headphones.

Now we receive the correct ITD ( encoded in the recording) which will not be possible in speakers playback because the ITD cannot exceed the delay produced by speakers angle.

I think there's some confusion here. The correct ITD is encoded in the recording only with binaural type of recordings. With traditional recordings, it is very likely that there's no readily interpretable binaural cues in the L-R signals.
The ITD between, for example, loudspeaker L signal as received from the left ear and the right ear has nothing to do with the encoded ITD in the recording itself, if done with a binaural microphone. The loudspeaker crosstalk ITD is pretty much constant, while the one in a binaural recording depends on the location of the instrument that's being recorded.

However with headphones you lose the role of pinna in direction finding which will cause the sound to be internalized. A sound is 3D when the ITD, ILD and pinna function are delivered in tact to the brain. A binaural recording sound 3D because it is recorded and reproduce at the same spots.

With headphones you don't necessarily lose directional cues. If they are encoded in the recording as HRTFs, the bypassing of the pinna is not a problem at all. If what you say about binaural recordings were true, you could get a full 3D experience by putting a spaced pair of microphones 60 deg apart to where you want the virtual soundstage to be represented in respect to.
But binaural works because it is recorded at very specific microphone capsule points, where the HRTFs are encoded on the recording for all directions of incoming sound.
After those cues are encoded, it is a matter of making sure that they are provided to the respective ear back in the playback stage.
With headphones you have no cross-talk problems, but you do have other headphone playback related issues. Namely, the difference in acoustical impedance seen by the eardrum when you put a cup over your pinna.
With loudspeaker you have a natural acoustical load seen by your eardrums, but you do lose some of the binaural cues.
Speaker cross-talk cancellation could help with that, if done correctly.
But that's the opposite perspective of what the OP was asking for, so probably off topic..

 

[STC]

I think there's some confusion here. The correct ITD is encoded in the recording only with binaural type of recordings. With traditional recordings, it is very likely that there's no readily interpretable binaural cues in the L-R signals.
The ITD between, for example, loudspeaker L signal as received from the left ear and the right ear has nothing to do with the encoded ITD in the recording itself, if done with a binaural microphone. The loudspeaker crosstalk ITD is pretty much constant, while the one in a binaural recording depends on the location of the instrument that's being recorded.

Unless you use XY microphones, most stereo microphones encoded the ITD depending on the spacing. Of course, for the correct ITD the spacing of the microphones must match your ear spacing.

With binaural recordings, the ITD matches your ear spacing. Even then it will not be accurate as our pinnae are unique for each individuals.

For an example, if you the sax is about 90 degrees to the left in live performance, the sound will reach your right ear by about 400 μs( you can calculate the correct delay). A stereo ( spaced AB technique) or binaural recording correctly will encoded the ITD.

Problem starts when you play them with loudspeakers at 60 degrees ( typical stereo setup). The sound of the left speaker will now reach your right ear by about 220 μs. This delay is cause by the physical sound source of the left speaker. And at the same time the right speaker will also launch the second signal encoded in the recording of the original 400 μs delay.

Now you brain is presented a first ITD due to physical location of 220 μs first before hearing the original encoded ITD coming from the right speakers which will arrive by another 180 μs later then the ITD caused by the physical speaker location. The brain will utilize the first 220 μs for localization and ignore the original 400 μs ITD. This explains why the sound width cannot extend beyond the physical boundary of the loudspeakers. Having said that, it is possible to utilize phase manipulation like QSound to place the sound beyond the physical outer boundaries of the speakers but that’s another topic.

In a binaural recording, you still will have the same error when you produce them over the loudspeakers. By using headphones, you eliminate the interaural crosstalk error so you perceive the encoded ITD and ILD with errors and the stage is much wider

With headphones you don't necessarily lose directional cues. If they are encoded in the recording as HRTFs, the bypassing of the pinna is not a problem at all. If what you say about binaural recordings were true, you could get a full 3D experience by putting a spaced pair of microphones 60 deg apart to where you want the virtual soundstage to be represented in respect to.
But binaural works because it is recorded at very specific microphone capsule points, where the HRTFs are encoded on the recording for all directions of incoming sound.
After those cues are encoded, it is a matter of making sure that they are provided to the respective ear back in the playback stage.
With headphones you have no cross-talk problems, but you do have other headphone playback related issues. Namely, the difference in acoustical impedance seen by the eardrum when you put a cup over your pinna.
With loudspeaker you have a natural acoustical load seen by your eardrums, but you do lose some of the binaural cues.
Speaker cross-talk cancellation could help with that, if done correctly.
But that's the opposite perspective of what the OP was asking for, so probably off topic..

Without pinna’s role, the sound will be internalized. The main difference between binaural and stereo recordings is the pinna frequencies shaping function is missing in stereo stereo recordings. Since a binaural recording is made using a dummy head or by the maker wearing a binaural microphones, the pinna function is engaged and frequency is shaped according the the direction where it is coming from. See Moller’s paper on the measurements of frequency response 360 degrees. The frequencies response difference gives the cues to the brain as to the direction where the sound originated from.

Sound will be 3Dimensional if you can deliver the ITD, ILD and pinna function reasonably close to how you hear real sound in nature. You may have heard of Smyth Realizer headphones which externalize the headphones sound.

 

[Deleted member 16543]

I somewhat agree with what you say in the first part. However, and I don't know why, when playing a binaural recording over loudspeakers I can still experience an expanded soundstage. Not as wide or 360 deg as with headphones, but personally it's a trade off I'm willing to make. I don't particularly like the sound of many headphones. I can venture to say that the later ITD is likely more close in frequency shaping to what the brain would expect, and that's probably what maintains some of the binaural sauce in the recording even over loudspeakers. Again, a well designed cross-talking cancellation algorithm could potentially make things a lot better, but it's tricky to get right. It could, for example, easily double the driver's excursion at lower frequencies in respect to the recording being normally played.

As for the pinna's role, I think we are saying the same thing. But it's not only the pinna role that matters in a binaural recording. It is the pinna, plus the head, plus the canal (plus the body for near sounds), as a whole system, that determines the difference in ITD, ILD and frequency shaping between left and right eardrum sound pressures. I actually wrote a compendium available on GS on the matter that says in much greater detail what I'm writing here. In short, the capture and the reproduction of a binaural signal are two completely separate things, the latter requiring an EQ to make up for the second pass over the listener's outer ear.
Headphones are supposed to sound not too different from loudspeakers (balance wise), and that's why they tend to be scooped around 3 kHz. They're undoing the frequency shaping of the pinna.
But I have to maintain that it is not the lack of a pinna that collapses the soundstage inside the head with a traditional recording. It is the inconsistency of the initial phase difference between L and R, which the brain doesn't know how to interpret and ends up resolving, virtually, inside the head. An example of the extreme opposite is the infinite piano effect, when the difference in phase of a note hitting the first spaced mic, then the second, is wider than what the head dimension would make it. So for phases smaller than the distance between ears the sound is place inside the head. For phases bigger than that, the sound is placed way farther to the side than it is in reality.
I wish I could find that video I was talking about. It is evident what I'm saying, there. Maybe I'll dig it up from somewhere tomorrow..

Either way, I think this thread is for the opposite task: adding electronic cross-talk to headphones to make them sound like speakers.
We are talking about eliminating acoustic cross-talk from loudspeakers to maintain the binaural cues as they are maintained with headphones.

 

[STC]

I somewhat agree with what you say in the first part. However, and I don't know why, when playing a binaural recording over loudspeakers I can still experience an expanded soundstage. Not as wide or 360 deg as with headphones, but personally it's a trade off I'm willing to make. I don't particularly like the sound of many headphones. I can venture to say that the later ITD is likely more close in frequency shaping to what the brain would expect, and that's probably what maintains some of the binaural sauce in the recording even over loudspeakers. Again, a well designed cross-talking cancellation algorithm could potentially make things a lot better, but it's tricky to get right. It could, for example, easily double the driver's excursion at lower frequencies in respect to the recording being normally played.

As for the pinna's role, I think we are saying the same thing. But it's not only the pinna role that matters in a binaural recording. It is the pinna, plus the head, plus the canal (plus the body for near sounds), as a whole system, that determines the difference in ITD, ILD and frequency shaping between left and right eardrum sound pressures. I actually wrote a compendium available on GS on the matter that says in much greater detail what I'm writing here. In short, the capture and the reproduction of a binaural signal are two completely separate things, the latter requiring an EQ to make up for the second pass over the listener's outer ear.
Headphones are supposed to sound not too different from loudspeakers (balance wise), and that's why they tend to be scooped around 3 kHz. They're undoing the frequency shaping of the pinna.
But I have to maintain that it is not the lack of a pinna that collapses the soundstage inside the head with a traditional recording. It is the inconsistency of the initial phase difference between L and R, which the brain doesn't know how to interpret and ends up resolving, virtually, inside the head. An example of the extreme opposite is the infinite piano effect, when the difference in phase of a note hitting the first spaced mic, then the second, is wider than what the head dimension would make it. So for phases smaller than the distance between ears the sound is place inside the head. For phases bigger than that, the sound is placed way farther to the side than it is in reality.
I wish I could find that video I was talking about. It is evident what I'm saying, there. Maybe I'll dig it up from somewhere tomorrow..

Either way, I think this thread is for the opposite task: adding electronic cross-talk to headphones to make them sound like speakers.
We are talking about eliminating acoustic cross-talk from loudspeakers to maintain the binaural cues as they are maintained with headphones.

Perhaps a new thread would be better option. I thought since this thread comes under room acoustics and general speakers, it can be useful for both headphones and loudspeakers listeners.

A binaural recording will sound too bright and weird over loudspeakers. Even the so called binaural + recordings of Chesky are not really binaural because the pinna correction was applied to them to make them to sound normal. Pinna alters the frequency response and go go up to 20dB at certain frequencies.

For a sound to be natural, it must be delivered with the ILD, ITD and pinna cues to the ears. A non binaural recording will sound binaural when you eliminate crosstalk. Listen to BACCH or Ambiophonics setup. But the stage is limited to 180 frontal stage. Since there is no height information encoded in stereo, you won’t hear height but for everyday music listening it is as good as live performance.

regarding the need for head, torso and others for binaural recording, please listen to recordings made with 3Dio microphones. There is no dummy head except for pinnae and yet it sounds very 3D.

If you start a new topic on this please tag me.
Cheers and thanks for the discussion.

 

[Inner Space]

For a sound to be natural, it must be delivered with the ILD, ITD and pinna cues to the ears.

These days DSP has a lot of power, and the headphone market is huge, so why isn't this happening? Imagine if you could put on a comfortable pair of headphones, with soft pads, and get lost in a huge acoustic universe, miles wide, miles deep, full of enchanting sounds. Surely such a thing isn't beyond modern programming talent?

I think it has happened, somewhere, but it didn't really work. The problem is we hear with more than the eardrum. Whole-body, chest, and certainly the bones of the skull. This is much more than collateral damage, left over from sound arriving at the ear. It's a big part of hearing. I love headphones, and use them part of the year when I'm in the city, but eardrum-only devices will never sound real.

 

[JoachimStrobel]

A reminder: Vinyl has a channel separation of 30db. A lot of talk about the so good sounding LPs seems to originate in that low separation, creating a blurred soundstage.

 

[Deleted member 16543]

Perhaps a new thread would be better option. I thought since this thread comes under room acoustics and general speakers, it can be useful for both headphones and loudspeakers listeners.

A binaural recording will sound too bright and weird over loudspeakers. Even the so called binaural + recordings of Chesky are not really binaural because the pinna correction was applied to them to make them to sound normal. Pinna alters the frequency response and go go up to 20dB at certain frequencies.

For a sound to be natural, it must be delivered with the ILD, ITD and pinna cues to the ears. A non binaural recording will sound binaural when you eliminate crosstalk. Listen to BACCH or Ambiophonics setup. But the stage is limited to 180 frontal stage. Since there is no height information encoded in stereo, you won’t hear height but for everyday music listening it is as good as live performance.

regarding the need for head, torso and others for binaural recording, please listen to recordings made with 3Dio microphones. There is no dummy head except for pinnae and yet it sounds very 3D.

If you start a new topic on this please tag me.
Cheers and thanks for the discussion.

I think you call binaural any synthesized, virtual spatially enhanced recording. I use that term specifically for the ACCURATE capture and reproduction of binaural cues as they would be sensed by the listener during the recording.
To me, synthesized binaural doesn't qualify for the term. The 3Dio doesn't qualify either, since it lacks the head shadow effect.
Ambisonics 'treated' stereo recordings definitely don't qualify. The 3D effect is random and depends on the randomly recorded phase information encoded in the stereo recoding typical of traditional multi mono to stereo mix down. Although I must admit they do sound pretty cool!
I think this is where the disconnect happens between us in terminology on this topic.

I'm more interested in making accurate speakers these days, but if you want a somewhat lengthy explanation of why you need at least the whole geometry of the head for accurate binaural recordings, this is that compendium I wrote a while ago.
https://www.gearslutz.com/board/all-things-technical/1288581-binaural-theory.html#post14337740

 

[Frank Dernie]

Late to this but:
About 20 years ago we experimented with a few of the shortcomings of record player systems to see why they didn't sound all that bad despite the awful specs. One of the tests was crosstalk.
A very good pickup cartridge has crosstalk in the 25 to 35 dB range ar mid frequencies. Neither of us listening could hear any difference when crosstalk was increased from the maximum in the digital file to 30dB. Completely inaudible to us both.

I used to frequently do my own recordings using a rented dummy head. The balance was achieved entirely by moving the "head" during rehearsals. I find these recordings very natural over headphones and speakers, though I must admit I am judging instrumental balance and timbre (very important to me) much more than imaging when I am listening.