What’s Your Triangle in Stereo Speaker Listening?

[Tim Link]

I use m/s plugin for the ambiance and did some experience by not cancelling the mono sound or add stronger centre using the M/S plugin. I think the same effect is included in some XTC DSP where it allows you to adjust the centre image strength.

3 channel stereo is at one time was popular and I think it was Bell which did or advocate this more than half a century ago. My curiosity is should the speakers angle extend beyond 60 or so degrees.

Bit off topic here below.

At one time, when Audio Note introduced their speakers it was setup about more than 120 degrees angle by one ‘audiophile expert’. Great reviews about the speakers and the expert’s skills in bring out the best. I must be the only one to say the sound is watered down although the stereo width was at 120 or so degrees but that doesn’t require special skills. Just record one sound in one channel each and you have the width proportional to the speakers placement angles. Some minimalist recordings like the Waterlily acoustics albums sounded nice but Tracy Chapman really sucked. But since the majority praised the system I knew I have serious taste and hearing problem.

The only reliable source why 60 degrees was chosen as the default position was the explanation offered by the eminent recording engineer Robin Miller where he said the mics were placed at 120 degrees and proper level is attained when the speakers angle are half of that.

Dolby research suggest 44 to 60 for the Atmos and it used to be up to 65 degrees before Atmos. Wilson Chronosonix suggest only about 43 degrees angle and so too Harbeth designers preference despite in the manual the suggestion was 60 degrees.

It is matter of taste and depends what your focus is at the time.

That's really interesting about the 'audiophile expert' setting up with a 120 degrees. I'm noticing some anomalies, like sounds moving with the direction I'm looking if I turn my head to look at the left or right speaker. For this reason the speakers only 'disappear' if I'm not looking right at them. If I turn my head far enought to look straight at one both ears can get a lock on its position.
Being suggestable, I just tested this wide setup on some Tracy Chapman and it sounded nice, easy to listen to and engaging. Maybe not jaw droppingly good but I wouldn't say it sucks. The 1988 mix of Fast Car is dramatically different than the 2015 mix on the Greatest Hits album. The live version of Stand By Me makes her voice sound great. I put on Meeting by the River to try some Waterlilly Acoustics. Sounds really pretty and I've basically got a sitar panned all the way to the right and a guitar all the way to the left. There is ambience but it floats around each instrument, doesn't fill the space in-between. Ah, now there's a drum right in the middle. It's growing on me by the minute.

This one sounds really great

I agree, it's largely a matter of taste and focus at the time. People should feel free to experiment. I'm glad I tried going wide. A standard width traingle is more stable. My experiment with adding mono in to narrow the soundstage has a sort of mixed result. The sound loses some impact and depth. The center gets louder while the edges get quieter, but they're actually still wide out there until they vanish below my hearing threshold. I tried moving the speakers to a narrower angle but the sound overall is not as good. So I guess I'll stick with this excessively wide soundstage for now. Maybe not accurate as intended in terms of soundstage width but wins overall. I'll keep thinking of ways to get rid of the phantom center crosstalk in a typical listening triangle. What we need is a way to employ XTC increasingly to center panned sounds, fading to zero as things are panned harder to the left or right. One way to potentially do that is with what I'm coining "personal acoustics." You can hug them like a teddy bear while you listen and they'll block the high frequencies from reaching the wrong ear but won't widen the soundstage significantly. I tried it and it works pretty well. I could even look over the top of it without it blocking my view of TV. It needed to come up to my chin to work well. After a while it made my chin itch.

 

[GoFaster]

That's really interesting about the 'audiophile expert' setting up with a 120 degrees. I'm noticing some anomalies, like sounds moving with the direction I'm looking if I turn my head to look at the left or right speaker. For this reason the speakers only 'disappear' if I'm not looking right at them. If I turn my head far enought to look straight at one both ears can get a lock on its position.
Being suggestable, I just tested this wide setup on some Tracy Chapman and it sounded nice, easy to listen to and engaging. Maybe not jaw droppingly good but I wouldn't say it sucks. The 1988 mix of Fast Car is dramatically different than the 2015 mix on the Greatest Hits album. The live version of Stand By Me makes her voice sound great. I put on Meeting by the River to try some Waterlilly Acoustics. Sounds really pretty and I've basically got a sitar panned all the way to the right and a guitar all the way to the left. There is ambience but it floats around each instrument, doesn't fill the space in-between. Ah, now there's a drum right in the middle. It's growing on me by the minute.

This one sounds really great

I agree, it's largely a matter of taste and focus at the time. People should feel free to experiment. I'm glad I tried going wide. A standard width traingle is more stable. My experiment with adding mono in to narrow the soundstage has a sort of mixed result. The sound loses some impact and depth. The center gets louder while the edges get quieter, but they're actually still wide out there until they vanish below my hearing threshold. I tried moving the speakers to a narrower angle but the sound overall is not as good. So I guess I'll stick with this excessively wide soundstage for now. Maybe not accurate as intended in terms of soundstage width but wins overall. I'll keep thinking of ways to get rid of the phantom center crosstalk in a typical listening triangle. What we need is a way to employ XTC increasingly to center panned sounds, fading to zero as things are panned harder to the left or right. One way to potentially do that is with what I'm coining "personal acoustics." You can hug them like a teddy bear while you listen and they'll block the high frequencies from reaching the wrong ear but won't widen the soundstage significantly. I tried it and it works pretty well. I could even look over the top of it without it blocking my view of TV. It needed to come up to my chin to work well. After a while it made my chin itch.

paragraphs are your friend

 

[STC]

For this reason the speakers only 'disappear' if I'm not looking right at them. If I turn my head far enought to look straight at one both ears can get a lock on its position.

Speakers disappear is a concept I don’t completely comprehend. Stereo creates phantom images between the speakers and technically the speakers disappear in such situation but when it comes with hard panned sound, it is now a single source and we localize them like all sound in nature.

Being suggestable, I just tested this wide setup on some Tracy Chapman and it sounded nice, easy to listen to and engaging. Maybe not jaw droppingly good but I wouldn't say it sucks.

Are you using centre channel?

Usually the bigger the soundstage is the intensity may suffer a little. Those familiar with wide stereo in the 70s would have noticed this phenomenon with the TV setting. Again this is all about preference unless we are obsessed with how much the level of , say , a guitar on the right compared to the centre vocal. The difference will change accordingly with the spread of the speakers.

I'll keep thinking of ways to get rid of the phantom center crosstalk in a typical listening triangle. What we need is a way to employ XTC increasingly to center panned sounds, fading to zero as things are panned harder to the left or right. One way to potentially do that is with what I'm coining "personal acoustics."

Theoretically, you lose 3dB due to XTC but with Ambiodipole placement ( that is closely placed speakers) you are actually reinforcing the mono sound but since your approach is different YMMV.

In any case, the correct position of stereo ( two speakers conventional placement already been researched and determined to about 45 to 60 degrees. Having said that taste difference and it too can change. I have forgotten how many times I said this is best setup until few months later then no no this is the best setup ( during my stereo era).

The wooden floor were full of spikes indentation of speakers varying speakers placement. I even experienced at 180 degrees!! Eventually, I found after two years I have been going in circles and the placement actually changes according to which track I am listening at that particular time.

cheers!

 

[goat76]

Usually the bigger the soundstage is the intensity may suffer a little. Those familiar with wide stereo in the 70s would have noticed this phenomenon with the TV setting. Again this is all about preference unless we are obsessed with how much the level of , say , a guitar on the right compared to the centre vocal. The difference will change accordingly with the spread of the speakers.

From the viewpoint of setting up a correct-sounding phantom center image (if that is the goal) while optimizing the positioning of the loudspeakers, I don't think you should rely on the comparisons of a recording you don't have full knowledge of how it is supposed to sound, as in your example with the comparison of the level of a right-panned guitar to centered-panned vocal.
Without a known reference of how that particular recording should sound like; as in not knowing the intended level differences (or not) between the guitar and the vocals, not knowing the intended perceived width of those individual sound objects, not knowing how distinct-sounding (or not) the two individual sound objects are meant to be sounding, and so on... you would end up optimizing the speaker positioning to how you would like that single particular mix to sound, which may or may not suit your preferences of how you would like the mix of the next track to sound.

My suggestion for setting up a correct-sounding phantom center image is instead to use a recording you have full controlled knowledge of. A recording of a sound object panned to different positions in the stereo field can be used as that sound object should have the same perceived level, the same perceived width, and sound about as "distinct" wherever that sound object is panned in the stereo field. When knowing that the sound object is the same without any differences in processing depending on the position in the stereo field, that sound should be perceived to sound about the same panned as a phantom-centered sound as it sounds hard-panned to either the left or the right loudspeaker. It should not sound "stretched out" or "washed out" in comparison to the hard-panned sound.

I have made a recording for this that you can use. It's a snare drum, a kick drum, and a human voice which are one at a time panned hard left, 50% to the left, phantom center, 50% to the right, and hard-panned to the right. As in the above description, these sound objects should sound about the same in level, width, and distinctiveness wherever they are panned in the stereo field.

Dropbox

www.dropbox.com

P.S. If you find that an equilateral triangle doesn't give you the correct level, width, and distinctiveness of the phantom center as the hard-panned sound, try to shrink the size of the triangle while still keeping it equilateral, as different types of loudspeakers depending on size and dispersion characteristics will likely dictate how far apart they can be positioned while still maintaining correct level, width, and distinctiveness of the phantom center image.

 

[STC]

From the viewpoint of setting up a correct-sounding phantom center image (if that is the goal) while optimizing the positioning of the loudspeakers, I don't think you should rely on the comparisons of a recording you don't have full knowledge of how it is supposed to sound, as in your example with the comparison of the level of a right-panned guitar to centered-panned vocal.
Without a known reference of how that particular recording should sound like; as in not knowing the intended level differences (or not) between the guitar and the vocals, not knowing the intended perceived width of those individual sound objects, not knowing how distinct-sounding (or not) the two individual sound objects are meant to be sounding, and so on... you would end up optimizing the speaker positioning to how you would like that single particular mix to sound, which may or may not suit your preferences of how you would like the mix of the next track to sound.

My suggestion for setting up a correct-sounding phantom center image is instead to use a recording you have full controlled knowledge of. A recording of a sound object panned to different positions in the stereo field can be used as that sound object should have the same perceived level, the same perceived width, and sound about as "distinct" wherever that sound object is panned in the stereo field. When knowing that the sound object is the same without any differences in processing depending on the position in the stereo field, that sound should be perceived to sound about the same panned as a phantom-centered sound as it sounds hard-panned to either the left or the right loudspeaker. It should not sound "stretched out" or "washed out" in comparison to the hard-panned sound.

I have made a recording for this that you can use. It's a snare drum, a kick drum, and a human voice which are one at a time panned hard left, 50% to the left, phantom center, 50% to the right, and hard-panned to the right. As in the above description, these sound objects should sound about the same in level, width, and distinctiveness wherever they are panned in the stereo field.

Dropbox

www.dropbox.com

P.S. If you find that an equilateral triangle doesn't give you the correct level, width, and distinctiveness of the phantom center as the hard-panned sound, try to shrink the size of the triangle while still keeping it equilateral, as different types of loudspeakers depending on size and dispersion characteristics will likely dictate how far apart they can be positioned while still maintaining correct level, width, and distinctiveness of the phantom center image.

I always used my own recordings since stereo days. Now I use my own recordings binaural with distinct sound every 10 degrees with varying distance. The level is determined by SPL meter and then compared the playback recording to see how accurate it replicate the original scene.

Now I just rely on AirPod Pro. as long as the playback sounds like what I hear with the AirPod Spatial I am happy.

Thanks for the link.

 

[Tim Link]

Are you using centre channel?

Usually the bigger the soundstage is the intensity may suffer a little. Those familiar with wide stereo in the 70s would have noticed this phenomenon with the TV setting. Again this is all about preference unless we are obsessed with how much the level of , say , a guitar on the right compared to the centre vocal. The difference will change accordingly with the spread of the speakers.

No, I'm not using a center channel with this wide setup. What surprised me about this is how good the phantom center sounds to me compared to a standard 60 degree triangle setup. It's very clear and intensity is excellent. I think it's because head shadowing is improved with the wider spacing so phantom center comb filtering at each ear is reduced above a certain frequency.

I have made a recording for this that you can use. It's a snare drum, a kick drum, and a human voice

I just listened to that. Thanks! It verifies what I've been hearing - that the phantom center sounds nearly identical in tone and level to the sides with this very wide setup. I never get that impression on a 60 degree triangle arrangement, and I don't see how it's possible unless the side panned sounds are intentionally time doubled to create a comb filter effect to match the comb filtering expected from the crosstalk of the phantom center in a 60 degree triangle. I don't know that anybody actually does that. That would be one way to get everything to sound the same without somehow addressing interaural crosstalk. I suspect it'd also not be a great sounding mix. The paradoxical thing about this is that 60 degrees is, in a way, about the worse case scenario for a degraded phantom center, and the 'sweet spot' is arguably the worse position for degrading the tone of the phantom center. Narrower spacing pushes the comb filtering up higher so it's not starting right inside the critical midband. Wider spacing can reduce the comb filtering by providing better head shadowing. Sitting off to one side allows one speaker to dominate, again reducing the comb filtering.

 

[STC]

No, I'm not using a center channel with this wide setup. What surprised me about this is how good the phantom center sounds to me compared to a standard 60 degree triangle setup. It's very clear and intensity is excellent. I think it's because head shadowing is improved with the wider spacing so phantom center comb filtering at each ear is reduced above a certain frequency.

At 60 degrees, say you sit 2.5m from the centre of the two speakers and the level of a sound from the left speaker is 80dB ( guitar) and the right speaker is also 80dB ( sax) . The distance from the speaker to the head is about 2.88m. When you move your speakers to 120 degree along the median plane and your distance now from speaker to head will be 5m. Since distance has increased the sound pressure would be dropping by 5 db. So the sound of the guitar and sax you heard at 80dB is now only 75dB.

And if that is a mono sound. At 60 degree, it might be slight above 80db due to doubling of the source . But at 120 degree, at best you can get is 78dB . So I just don’t understand why you perceive it the way you describe. of course, you can always increase the level but still you would notice the centre will be at least 2dB lower relatively compared to the 60 degrees setup.

 

[tmuikku]

center is product of the two speakers, so it also drops. And, early reflections increase as in reoation! But, it's much more easier to move listeners bottoms closer to speakers, than move speakers further apart. Also, early reflections reduce relative to direct sound takes no more than few seconds to move the listener and makes AB testing possible, just move yourself a bit. Put chair suitable position, lean back, or lean forward and even the bottoms don't have to be moved.

 

[Tim Link]

At 60 degrees, say you sit 2.5m from the centre of the two speakers and the level of a sound from the left speaker is 80dB ( guitar) and the right speaker is also 80dB ( sax) . The distance from the speaker to the head is about 2.88m. When you move your speakers to 120 degree along the median plane and your distance now from speaker to head will be 5m. Since distance has increased the sound pressure would be dropping by 5 db. So the sound of the guitar and sax you heard at 80dB is now only 75dB.

And if that is a mono sound. At 60 degree, it might be slight above 80db due to doubling of the source . But at 120 degree, at best you can get is 78dB . So I just don’t understand why you perceive it the way you describe. of course, you can always increase the level but still you would notice the centre will be at least 2dB lower relatively compared to the 60 degrees setup.

It's comb filtering. Both speaker's sound end up reaching both ears, but at slightly different times. It's already been shown earlier in this thread that you end up with a dip in response around 2K for the phantom center even with room reflections helping to fill in. That's a little misleading because the inital sound that reaches your ears matters, and with this cross talk across the head it's like a very strong and very early reflection. So room reflections help to mask it, but they don't hide it. To play the speakers wider you can move them closer so the distance is the same as before. In the test posted earlier I percieved the center as maybe slightly louder than the side panned sounds because I could hear it clearly in both ears simultaneously, while when panned I can only hear it loudly in one ear or the other.

But, it's much more easier to move listeners bottoms closer to speakers

Yes, a better direct to reflected ratio will help you hear more clearly, but it will actually accentuate the comb filtering caused by the phantom center because you won't have comb filtering from room reflections masking it as much. No matter how close or far you are from the speakers, crosstalk remains a problem based on the listening angle. I know a lot of people don't seem to perceive it, and perhaps they are lucky! I'm not saying the phantom center sounds aweful. I've heard it sound very good in a normal setup. It just doesn't sound as good as it potentially could unless something is done about the crosstalk. That requires either physically blocking some of the crosstalk with a barrier, or using inverse, time delayed signals to recursively cancel it. With a wide speaker spacing, you make your own head into a more effective barrier, and you get improved phase differential across the head in the lower frequencies.

 

[krabapple]

I've only seen the term once on this thread , but aren't we mainly talking about near vs far field listening?

 

[tmuikku]

It's comb filtering. Both speaker's sound end up reaching both ears, but at slightly different times. It's already been shown earlier in this thread that you end up with a dip in response around 2K for the phantom center even with room reflections helping to fill in. That's a little misleading because the inital sound that reaches your ears matters, and with this cross talk across the head it's like a very strong and very early reflection. So room reflections help to mask it, but they don't hide it. To play the speakers wider you can move them closer so the distance is the same as before. In the test posted earlier I percieved the center as maybe slightly louder than the side panned sounds because I could hear it clearly in both ears simultaneously, while when panned I can only hear it loudly in one ear or the other.


Yes, a better direct to reflected ratio will help you hear more clearly, but it will actually accentuate the comb filtering caused by the phantom center because you won't have comb filtering from room reflections masking it as much. No matter how close or far you are from the speakers, crosstalk remains a problem based on the listening angle. I know a lot of people don't seem to perceive it, and perhaps they are lucky! I'm not saying the phantom center sounds aweful. I've heard it sound very good in a normal setup. It just doesn't sound as good as it potentially could unless something is done about the crosstalk. That requires either physically blocking some of the crosstalk with a barrier, or using inverse, time delayed signals to recursively cancel it. With a wide speaker spacing, you make your own head into a more effective barrier, and you get improved phase differential across the head in the lower frequencies.

Yeah crosstalk gets very obvious listened with reduced early reflections, and as a side product makes the equidistant center, the listening spot, easy to find. This combfilter has fun feel to it though, it feels like descent into the recording as I physically approach the listening spot and I know exactly when I'm landed as the combfilter guides me to the spot. While it's not an ideal thing it's part of the charm of stereo I think, part of the magic trick.

In your previous post you mentioned head shadowing and I had just played around with it bit earlier today. It somehow makes the highs feel calm. I mean, put mono pink noise playing from both speakers, go listen on to your sweetspot but the highs could be kind of obvious on top of the noise, attention grabbing. Move a bit closer so angle to speakers increase and the feel suddenly disappears, as if the highs blend better in to the noise. If I turn my head, look at either of the speakers the HF noise comes back in to perception, which indicates it's the shadowing, or the combfilter. Now, this might be just how my system voicing is setup but it is kind of interesting thing to experiment with and I feel your appeal to the crosstalk cancellation thing. Stereo sound has so many facets to it we likely wonder about this stuff years to come

 

[STC]

Both speaker's sound end up reaching both ears, but at slightly different times.

Yes but when you start hearing a sound longer you perceive it to be louder. This is different from putting UMIK mic in the centre and do the measurement. I don’t know where you get the dip but using Moller research for sound coming 60 degrees angles you have 10 dB attenuation in the opposite ear at 2 kHz due to head shadow and for 120 degrees placement it will be about 16 dB attenuation for 2kHz. Again, the level is significantly perceived lower due to wider placement.

Anyway, I was asking about the the loss of level due to increased distant unless you 120 degrees is now in equal distance like the 60 degrees by moving them in semi circle but then the head shadow attenuation is still more.

 

[LIΟN]

interesting content
We'll be able to recognize that by ear measurements.
But the characteristics of the angle and ITD and ILD
It can also be determined by the anechoic chamber dummy head 3m distance.

30 degree 229.2us ITD

60 degree 458.3 ITD

each ILD normalized

Since we listen to both together, 30/60 angle total phase comparison of the total response.

Comparison of the ExcessPhase without the influence of frequency(minphase) amplitude

Sum Response(tonal balance when we listen in realilfe each degree)

 

[STC]

@Lion, thanks for the measurements. Your is about 10 db difference between 30 and 60 degrees.

Since you have the facilities to do in ear measurement, try measuring two successive signal delayed by 60μs , 220μs and 440μs. I am just curious to see the measured frequency response difference.

Thank you.

 

[LIΟN]

thanks for the measurements. Your is about 10 db difference between 30 and 60 degrees.

Since you have the facilities to do in ear measurement, try measuring two successive signal delayed by 60μs , 220μs and 440μs. I am just curious to see the measured frequency response difference.

You're very welcome. Like you, I could also attach data from my own in-ear measurements, but I’ve shared a more controlled response from a public anechoic chamber dummy head instead.

Are you asking about the responses at different angles with the three delay times you mentioned, each corresponding to different ITDs?
Or would you like me to simulate and adjust the delay (60μs, 220μs, 440μs) and observe how it alters the combined response? ---> Such as 30degree has 229.2us itd, and change to 60us? or 440us?
If it's the latter, the inherent ITD of the response would be disrupted, so I'm not sure if the results would be meaningful.
It seems like you're probably referring to the former case, but it's a bit unclear. (It could also be that my translator is interpreting it strangely.)

 

[STC]

You're very welcome. Like you, I could also attach data from my own in-ear measurements, but I’ve shared a more controlled response from a public anechoic chamber dummy head instead.

Are you asking about the responses at different angles with the three delay times you mentioned, each corresponding to different ITDs?
Or would you like me to simulate and adjust the delay (60μs, 220μs, 440μs) and observe how it alters the combined response? ---> Such as 30degree has 229.2us itd, and change to 60us? or 440us?
If it's the latter, the inherent ITD of the response would be disrupted, so I'm not sure if the results would be meaningful.
It seems like you're probably referring to the former case, but it's a bit unclear. (It could also be that my translator is interpreting it strangely.)

Please see email. I don’t want to hijack this thread by going off topic.

 

[j_j]

You're very welcome. Like you, I could also attach data from my own in-ear measurements, but I’ve shared a more controlled response from a public anechoic chamber dummy head instead.

Are you asking about the responses at different angles with the three delay times you mentioned, each corresponding to different ITDs?
Or would you like me to simulate and adjust the delay (60μs, 220μs, 440μs) and observe how it alters the combined response? ---> Such as 30degree has 229.2us itd, and change to 60us? or 440us?
If it's the latter, the inherent ITD of the response would be disrupted, so I'm not sure if the results would be meaningful.
It seems like you're probably referring to the former case, but it's a bit unclear. (It could also be that my translator is interpreting it strangely.)

What you want to do is calculate the results at one ear for the linear sum of the two HRTF's corresponding to +- 30 INCLUDING itd. Observe both the confusion in arrival as well as the dip due to HRTF and ITD. For this, actual ITD's are important, because ITD isn't actually the same at every frequency for a given angle.

 

[LIΟN]

What you want to do is calculate the results at one ear for the linear sum of the two HRTF's corresponding to +- 30 INCLUDING itd. Observe both the confusion in arrival as well as the dip due to HRTF and ITD. For this, actual ITD's are important, because ITD isn't actually the same at every frequency for a given angle.

I almost know what you're talking about, but the precise comparison by frequency seems to be beyond what I can do. =(

i can arbitrarily check by hanging a 50hz IR filter.
The impulse at 50 Hz is very large, so you have to check it a little smaller.

------Wrong Align Delete----------

 

[LIΟN]

every frequency

I overlapped the filtered IR wrong earlier. I'm sorry.

50Hz Filterd IR Impulse
And When i use Align tool in REW

I got 0.39ms.

And In 60degree 50hz 0.67ms

 

[j_j]

The thing to compare is the resulting joint HRTF from both left and right to one ear (either one) to the actual zero degree hrtf.

It's an interesting indictment for the phantom center.