To crossfeed or not to crossfeed? That is the question...

[71 dB]

Ignoring what people say over and over and replying over and over without ever addressing their points doesn't earn respect.

I have tried to address those things from my perspective. I'm sorry my answers are not what you expect to see.

 

[gregorio]

I mean the ITD is not the same in your listening room it is in the mixing room unless the rooms are identical.

The Interaural Time Difference is the same, the distance between your ears doesn’t change when you listen in a different room but the sound entering your ears is obviously not the same.

For example if your room is bigger and less treated acoustically, the reflections will come to your ears later and with different level/spectrum/angle than in the studio.

Yes of course, you’re going to get a significantly different set of reflections, more early reflections, more time delay between the ERs, more spectral interaction between the ERs and between the direct sounds, longer RT60, etc. In addition, there’s going to be differences in the direct sound reproduced to start with, the different speakers in the two listening environments are going to have a somewhat different spectral/freq response and a somewhat different time domain response (group delay, etc.).

If that doesn't affect spatiality, I don't know what does.

Yes, it does of course affect spatiality but this is the spatiality of the sound that is entering the ears and then of course that sound interacts with the ears; the pinnae, the skull, body, etc.

One room gives nice bass, another room has good diffuse airy treble etc. This is the origin of my attitude of "omitting" factors.

If we could not hear the difference between those two different rooms/speakers/presentations that would be a justification for omitting those factors. That is why your reason for omitting factors doesn’t make any sense to me because obviously we can hear the difference between the two different rooms/presentations.

The studio where I had the mixing course had near and far field speakers and the difference between them was quite dramatic!

Exactly but this effectively contradicts your theory that “Not omitting them [those factors] can lead to extremely difficult and costly measures to control them for relatively minor benefits.” - You said above the difference “was quite dramatic!” but here you’re saying “relatively minor benefits”. Obviously we have a lot of factors involved here, not just the factors you omitted, but your rationale for omitting factors doesn’t correlate with your own observation and that was in exactly the same room just with a different speaker presentation.

I believe so too, hence the mentality of omitting these factors.

You are taking one example/property of human perception and applying it to a different context (headphone use). There’s two problems with this, either of which on their own can invalidate your “mentality of omitting these factors”: Firstly, you are ignoring other examples/properties of human perception when listening to speakers and secondly, you don’t have any reliable evidence that the different context (headphone use) doesn’t affect any of these factors.

Floyd Toole (and others) didn’t only demonstrate that the brain can adapt over time/training to certain weaknesses in frequency response, he demonstrated a great deal more, such as the importance of time domain performance, off-axis and other speaker/room performance issues. A good practical example of this is the old Yamaha NS10 phenomenon (see this article and accompanying research). In this case we’ve effectively got the same room and even the same presentation (near field monitors), the only difference is the actual speakers. Why, when there were dozens of different near field monitors available, did virtually all commercial studios in the world have NS10s and more interestingly in the context of this thread, why does it elicit more polarised opinion than pretty much any other “industry standard”? The freq response was poor compared to other near fields but what set it apart was it’s time domain response, it’s group delay/impulse response.

I believe it is not much different with headphones without acoustic environment.

What basis, apart from your personal perception, do you have for that belief? In fact your belief is contrary to a considerable amount of reliable evidence. In an acoustic environment we have speakers in a room and a considerable amount of resultant spatial/acoustic information, however, all this information correlates to our sense of sight. We see the speakers and the room and the spatiality/acoustic information we hear obviously correlates with that. Furthermore, when we listen in such an environment, we don’t have our head clamped in a vice, it’s moving around at least slightly and the resultant slight (or significant) changes in ITD and other factors reinforces the location/s of what we’re seeing (and hearing). And lastly, reliable evidence demonstrates that sight significantly influences positional hearing/perception. We don’t have any of this with headphones (unless they have HRTFs, head tracking and a reverb applied). And continuing:

The phase difference is too small to create comb-filter effects.

How do you know? Firstly, phase differences as low as 100micro-secs can cause audible comb-filter effects, although probably not in the lower freq band affected by crossfeed. Secondly, even though the phase difference is too small to create audible comb-filter effects in the lower band with coherent test signals, how do you know that’s the case with a stereo music mix, which is almost never phase coherent to start with, contains all sorts of direct mono and stereo sound sources, reflections and all sorts of audio effects, some/many of which are already partly out of phase to start with. Overlaying that and adding a further delay with crossfeed can indeed cause comb-filter effects or any similar/related type of effect (Doppler, flanging, phasing, etc.) and although not extreme, I’ve certainly perceived such effects when using crossfeed.

In addition, we’re not just talking about the spectral side effects of crossfeed delay but also the positional perception factors, and reliable evidence demonstrates that differences of as little as 5micro-secs can affect location perception. So for example, a sound in the mix with say a 600Hz fundamental will have that fundamental crossfed and delayed but it’s 2nd, 3rd and subsequent harmonics will not be, so you could perceive different spectral parts of the same sound to be in somewhat different locations.

With headphones, we’ve got no visual reference to influence/correct our perception of location perception and we’ve got the added complexity that the signals we’re listening to (stereo music mixes) not only are not spatially consistent/coherent internally but also do not correlate acoustically to our listening environment. The evidence indicates that under such conditions of limited sensory references (and those we do have conflict), plus the complex, confusing and contradictory aural cues within the music mixes themselves, results in our perception effectively having little to rely on and simply making-up whatever seems to make the most sense. This is why there is such a wide variety of individual responses to headphone listening from “sounds just like being there” to “a bit strange but I still like it” to “this is nonsense and very annoying”. Research is relatively limited in this case, it’s so far been mainly limited to understanding basic processes, such as location perception of single simple test signals, not multiple complex sounds with different locations and different acoustic information all occurring simultaneously. But, we can’t rationally just discount/ignore much of what we have discovered simply on the basis of one person’s perception, especially as it’s not representative of the majority.

G

 

[71 dB]

The Interaural Time Difference is the same, the distance between your ears doesn’t change when you listen in a different room but the sound entering your ears is obviously not the same.

Interesting how "different factors" are a serious thing with crossfeed: ITD is ruined just like that, but with room acoustics everything is okay. I brought this ITD thing with room acoustics because you have been "teaching" me not to omit "factors".

I am afraid I can't agree with you about ITD staying the same. The room dimension (+ speaker/listener positions) determine the angles the reflected sound arrives to the listeners ears. If the room is wider, the reflection from side walls not only arrive later compared to direct sound, but also at a larger angle generating bigger ITD. Sure the differences aren't massive, but they are there.

Yes of course, you’re going to get a significantly different set of reflections, more early reflections, more time delay between the ERs, more spectral interaction between the ERs and between the direct sounds, longer RT60, etc. In addition, there’s going to be differences in the direct sound reproduced to start with, the different speakers in the two listening environments are going to have a somewhat different spectral/freq response and a somewhat different time domain response (group delay, etc.).

Yep. The differences are many. It is very complex. Despite of this people "bite the bullet." Crossfeed makes consistent simple predictable modifications to the sound. I wonder which one is less problematic?

Yes, it does of course affect spatiality but this is the spatiality of the sound that is entering the ears and then of course that sound interacts with the ears; the pinnae, the skull, body, etc.

Using over-the-ear headphones the sound interacts with the pinnae (althou not the way it does with speakers because the sound is not arriving at different directions, but at one fixed direction, the headphone driver). Crossfeed approximates very roughly the interaction with skull (whereas not using crossfeed or anything else simulates situation where your head is inside a wall so that your ears/speakers are in different acoustically isolated rooms). Body is not simulated.

If we could not hear the difference between those two different rooms/speakers/presentations that would be a justification for omitting those factors. That is why your reason for omitting factors doesn’t make any sense to me because obviously we can hear the difference between the two different rooms/presentations.

How is hearing differences between no crossfeed and crossfeed any different? How am I omitting factors with crossfeed if I am not omitting factors when listening to speakers in my room?

Exactly but this effectively contradicts your theory that “Not omitting them [those factors] can lead to extremely difficult and costly measures to control them for relatively minor benefits.” - You said above the difference “was quite dramatic!” but here you’re saying “relatively minor benefits”. Obviously we have a lot of factors involved here, not just the factors you omitted, but your rationale for omitting factors doesn’t correlate with your own observation and that was in exactly the same room just with a different speaker presentation.

Benefits can still be minor while differences are dramatic. Both set of speakers gave good representation of the music in different ways. It is differing set of pros and cons.

You are taking one example/property of human perception and applying it to a different context (headphone use). There’s two problems with this, either of which on their own can invalidate your “mentality of omitting these factors”: Firstly, you are ignoring other examples/properties of human perception when listening to speakers and secondly, you don’t have any reliable evidence that the different context (headphone use) doesn’t affect any of these factors.

This is how my mind works, I guess... ....As an INTJ intuition is big part of how I "process" information. My senses have less to do with it.

This is all for now... I comment of the rest of your post later...

 

[gregorio]

Interesting how "different factors" are a serious thing with crossfeed: ITD is ruined just like that, but with room acoustics everything is okay.

The different factors are always a serious thing. Remove, ignore or even change one a bit and it can change/alter the perception. Everything isn’t OK with room acoustics but we have a great deal more spatial information, reflections/harmonics, more opportunity for masking and of course the mix and master are made by engineers for this sound presentation.

I am afraid I can't agree with you about ITD staying the same.

How can you not agree? The only way ITD can change is if the distance between the ears change. You think maybe when you listen in a different room your skull expands or contracts by several centimetres?

It is very complex. Despite of this people "bite the bullet." Crossfeed makes consistent simple predictable modifications to the sound.

And that’s the problem! It is very complex but it’s what our hearing and perception have evolved to deal with and it’s what we hear all day, every day for all our lives is also very complex. Crossfeed on the other hand is not very complex, HRTFs (head tracking and reverb) are far more complex and far better/closer to what our perception requires/expects. Crossfeed “makes consistent simple predictable modifications to the sound” mathematically but it’s obviously NOT predictable to most people’s perception because what you end up with when you apply crossfeed to complex musical mixes is various phase and other artefacts which are not similar to what happens with real sound in a real acoustic.

I wonder which one is less problematic?

That really should be obvious by now! It’s the simpler one which produces the spatial/spectral effects not expected/required by human perception.

You seem to still be missing the basic facts: We have a pair of complex signals which contains complex and somewhat contradictory spatial information, made more complex by speaker/room reproduction and we have hearing perception which uses a whole bunch of timing and spectral factors to make sense of what we’re hearing. Using headphones which don’t have that speaker/room interaction or other factors (such as head tracking) obviously lacks some of those factors which leaves perception with a lot more guesswork and therefore the results are unpredictable. Some people perceive a completely natural/realist result. Others perceive serious problems and not at all a natural/realistic result. Most though perceive something in between these two extremes, for example some (like me) perceive a reasonably natural result but from a much closer perspective and with a few artefacts/problems which aren’t a big deal. With crossfeed, we are still missing those factors and so our perception is still reliant on guesswork, however, the factors we do have are altered, some are improved and some degraded and this affects the result of our perception’s guesswork. For some people it completely ruins what was a very good/realistic result, for some (like you apparently) it completely cures what was an extremely poor result but again, most people fall between these two extremes. Some of this majority feel on balance that crossfeed has problems but is generally preferable and use it, most others feel on balance it’s not worth it, for example (like me) that it’s maybe a bit better with a few pieces but worse in most cases.

G

 

[71 dB]

The different factors are always a serious thing. Remove, ignore or even change one a bit and it can change/alter the perception. Everything isn’t OK with room acoustics but we have a great deal more spatial information, reflections/harmonics, more opportunity for masking and of course the mix and master are made by engineers for this sound presentation.

Why do you have to be so fast? I am not even finished with your prior post! I am so tired. I can try, but you have excuses (masking this time!)

How can you not agree? The only way ITD can change is if the distance between the ears change. You think maybe when you listen in a different room your skull expands or contracts by several centimetres?

ITD depends on the angle of the sound. That's the point of ITD! Different size room gives different angle for reflections => different ITD.

That's all for now.

 

[bigshot]

Even a bad room adds spatiality. Just the wrong kind. And it’s primarily a matter of timing reflections and delays.

 

[71 dB]

And that’s the problem! It is very complex but it’s what our hearing and perception have evolved to deal with and it’s what we hear all day, every day for all our lives is also very complex. Crossfeed on the other hand is not very complex, HRTFs (head tracking and reverb) are far more complex and far better/closer to what our perception requires/expects. Crossfeed “makes consistent simple predictable modifications to the sound” mathematically but it’s obviously NOT predictable to most people’s perception because what you end up with when you apply crossfeed to complex musical mixes is various phase and other artefacts which are not similar to what happens with real sound in a real acoustic.

Complex but natural! That's the key thing. You don't get crazy ILD at low frequencies with speakers in a room. Room "regulates" spatial cues so that they are VERY complex, but the spatial cues live in the natural range so that spatial hearing is able to understand it. With headphones the spatial cues are simpler, but they can be very unnatural which is the problem. Sure, crossfeed doesn't give the same things as room, but the it is more natural and that is the key for me.

 

[71 dB]

The low+high thing: Above about 1600 Hz ITD loses meaning, so it doesn't matter much if delay low vs high is different. At least I don't have issues, on the contrary wihtout crossfeed localization is a mess! Much better with crossfeed!

 

[bigshot]

No time change, no spatiality. Reverb adds spatiality.

 

[gregorio]

ITD depends on the angle of the sound. That's the point of ITD! Different size room gives different angle for reflections => different ITD.

Oh dear! Yes, ITD depends on the angle of the sound. So with a sound directly in front of us the ITD is zero, as the sound hits both our ears at the same time. Room size makes NO difference, whether the sound is 2m away or 20m, it still hits both ears at the same time and there’s zero ITD. Room size also makes no difference to angle, what makes a difference to angle is the relative position of the sound source (speaker for example) relative to the head position of the listener. Room size obviously makes a difference to the initial delay of the first early reflection and the delay of subsequent reflections but OF COURSE, that’s the initial delay/s of the early reflections and not ITD!!

Complex but natural! That's the key thing. You don't get crazy ILD at low frequencies with speakers in a room.

That is correct but also very incorrect. Yes, complex but natural is the key thing but there is NOTHING natural about only reducing ILD without also having room reflections, varying ITD and varying spectral spectral content of both the direct sound and reflections from both the room and the pinnae, skull, etc.

Sure, crossfeed doesn't give the same things as room, but the it is more natural and that is the key for me.

You’re contradicting yourself again. Crossfeed is not “more natural”, where do you get that from, it’s nonsense but still you keep repeating it! What is natural about a fixed ITD below a certain frequency, about a spectral effect/absorption that does not vary according to angle? All of this is completely unnatural. The only factor out of many that is natural with crossfeed is ILD, all the other factors are unnatural and you can’t simply say “I’m right because I ignore all those other factors” because it is contrary to the science and indeed to common sense, because if you were right there would be no need or point of developing HRTFs!

G

 

[71 dB]

Oh dear! Yes, ITD depends on the angle of the sound. So with a sound directly in front of us the ITD is zero, as the sound hits both our ears at the same time. Room size makes NO difference, whether the sound is 2m away or 20m, it still hits both ears at the same time and there’s zero ITD. Room size also makes no difference to angle, what makes a difference to angle is the relative position of the sound source (speaker for example) relative to the head position of the listener. Room size obviously makes a difference to the initial delay of the first early reflection and the delay of subsequent reflections but OF COURSE, that’s the initial delay/s of the early reflections and not ITD!!

When I talked about direct sound from speakers, you told me to not forget the other stuff, reflections and reverberation. Now you are talking about direct sound! I meant of course early reflections from side walls, and yes, room dimensions do affect the sound angle. It is basic geometry! Since you want to talk about direct sound, there is acoustic crossfeed happening with it. I have tried to make that point, but you constantly move the goalposts!

 

[bigshot]

With stereo speakers, the position of the speakers is a triangle with the listening position. As the room size increases, the triangle scales up so the sound is always coming from the same angles.

 

[71 dB]

With stereo speakers, the position of the speakers is a triangle with the listening position. As the room size increases, the triangle scales up so the sound is always coming from the same angles.

The shape of the room affects and also the distance to the side walls. Most of the time the angles* are quite similar, but if we are worried about 5 µs difference, those are certainly possible.

* If the side wall was all mirror, the angles are the ones that you see your spakers on the mirror. If you make the listening triangle smaller, the angle increases and vice versa. Theoretically it can be between 30° and 90°, but in practice it is about 50°-70°

 

[bigshot]

The angle of the sound from the speakers doesn’t change as the room size gets larger.

 

[71 dB]

You seem to still be missing the basic facts: We have a pair of complex signals which contains complex and somewhat contradictory spatial information,

With speakers this contradictory spatial information isn't a huge issue, because the room acoustics shape it into something less contradictory. With headphones that doesn't happen.

made more complex by speaker/room reproduction and we have hearing perception which uses a whole bunch of timing and spectral factors to make sense of what we’re hearing.

Naturally spatial sound has got so much information (all the reflections etc.), that I believe the analyse of it all in our brain happens largely statistically. It means that individual reflections (apart from direct sound + early reflections that are more sparse and can be analysed more carefully) don't matter so much, but the combination of all of them.

Using headphones which don’t have that speaker/room interaction or other factors (such as head tracking) obviously lacks some of those factors which leaves perception with a lot more guesswork and therefore the results are unpredictable.

Yes, and the result of this is I get broken messy spatiality with headphones as they are (without crossfeed) unless it is binaural or binaural-like sound.

Some people perceive a completely natural/realist result.

I don't and I don't understand how some people do. Of course this is a nice talent to have. Makes headphone listening simpler.

Others perceive serious problems and not at all a natural/realistic result.

That's me.

Most though perceive something in between these two extremes, for example some (like me) perceive a reasonably natural result but from a much closer perspective and with a few artefacts/problems which aren’t a big deal.

How do you hear large ILD at low frequencies? How does sound in only one ear feel to you? To me these things sound very unnatural and annoying. Using phone is annoying too, but thanks to the lack of low frequencies it is not very bad.

With crossfeed, we are still missing those factors and so our perception is still reliant on guesswork, however, the factors we do have are altered, some are improved and some degraded and this affects the result of our perception’s guesswork.

Things are altered in a way that roughly simulates how they are altered in regards of direct sound in "real life". So, even if some things degrade, they degrade similarly to "real life" that is familiar to my spatial hearing.

What is important to realize is that in stereo sound we can have channel differences, that don't really make sense to spatial hearing. It is like video signal with color information in IR and UV range that messed up the RGB color information. For example stereo signal where left and right channels are negative versions of each other doesn't make sense for human spatial hearing which expects certain types of cross-correlation between the channels. Mono sound (left and right channels identical) don't make sense in the context of spatial hearing either. When there is some differences, but not huge differencies is the "sweet spot". Room acoustics transform even the craziests ping pong recordings and even mono recordings to this sweet spot. Crossfeed does similar thing, but unfortunately it can't improve mono sound. For that I have developped "diffuse mono", where an artificial "S" channel is created from the original mono sound and delayed randomly before "added" to the original sound. That makes the mono sound diffuse. Mono recordings with headphones sound very "dead" and totally centered inside head. Making the mono diffuse makes it feel more outside head and also more lively, like a stereo recording where every instrument happened to be in the center. This is kind of the opposite of what crossfeed does (channel separation is increased instead of reduced). Unfortunately I can only do this processing beforehand, not "on the go".

For some people it completely ruins what was a very good/realistic result, for some (like you apparently) it completely cures what was an extremely poor result but again, most people fall between these two extremes.

It is interesting these differences exist among people, something I would have never believed before you educating me!

Some of this majority feel on balance that crossfeed has problems but is generally preferable and use it, most others feel on balance it’s not worth it, for example (like me) that it’s maybe a bit better with a few pieces but worse in most cases.

Maybe we simply have different threshold points of crossfeed being beneficial? For me crossfeed is beneficial for about 98 % of stereo recordings while for the rest 2 % it is not (for headphone compatible recordings crossfeed can be quite harmful, but those recordings are rare). For you it is the opposite?

Related to this my newer theory is that intuitive people favor crossfeed more than sensitive types. Intuitive types (MBTI = xNxx) suffer often from overwhelming sensory information while sensitive types (MBTI = xSxx) control sensory information better. Also, intuitive types process incoming information in order to find contradictions while sensitive types take information more as it is.

This is just an idea I have and can be totally wrong...