Wesayso,
This is where the idea of stereo two channel sound falls down, the idea of a phantom image from two distinct directions trying to simulate a direct sound from a specific direction. Out ears and brain are wired for reality and then we throw in this false image and expect it to fool us. A three channel system would have been much more realistic in my eyes but even that our ear/brain system would surely be able to detect the difference between that and a true point source real world. You just can't fool our evolved brain that way. We have been in a two channel music reproduction world for many years and have come to accept that, but it does not mean that it will ever fool the listener who is paying attention to details and comparing them to a live event.
You have done an excellent job at recreating the illusion that we have come to expect from the best of two channel sound systems. there is a limit to what we can do with that. You are now playing at the margins of possibility with that two channel system.
This is where the idea of stereo two channel sound falls down, the idea of a phantom image from two distinct directions trying to simulate a direct sound from a specific direction. Out ears and brain are wired for reality and then we throw in this false image and expect it to fool us. A three channel system would have been much more realistic in my eyes but even that our ear/brain system would surely be able to detect the difference between that and a true point source real world. You just can't fool our evolved brain that way. We have been in a two channel music reproduction world for many years and have come to accept that, but it does not mean that it will ever fool the listener who is paying attention to details and comparing them to a live event.
You have done an excellent job at recreating the illusion that we have come to expect from the best of two channel sound systems. there is a limit to what we can do with that. You are now playing at the margins of possibility with that two channel system.
Last edited:
I think you're right Steven.
Just for fun I had another look at Prof. Edgar Choueiri's impulses. I've linked the plots in the Phantom thread.
So I edited my chain, to mimic what was done there. Talk about a 3D stage! I only treated the mid component of mid/side processing. But the hard part is the tonality gets all messed up. Though with a lower inverted signal you can get some of it back to "normal".
What I did was create a center sum signal, invert it and band pass it and next combine it with the mid signal of my chain, about 17 dB below the original signal. It has the wanted effect of breaking loose the barriers of the sound, but it also alters perceived tonality. I immediately fired up REW to look at the signal sum at the ears, it still has cross talk dips, but the first wave hitting the ear now is an inverted image of that Dirac pulse. It really works though.
It filled my room out nicely, stage almost all around me basically. Now I've had that before, but this time the center played along. It is fun, but do I want to keep it? I may steel some ideas from it for a more toned down version.
Hard to say what route would be best, but I agree... a center channel would have been a smarter solution in the end.
If I get some free time I'll show some graph's of what's happening with the inverted phantom signal. It is fun as a learning exercise though. Prof. Choueiri's impulses finally make sense to me, even the wavy phase part of it. And it's easy to re-create in JRiver.
Got to do some more thinking I guess
.
Just for fun I had another look at Prof. Edgar Choueiri's impulses. I've linked the plots in the Phantom thread.
So I edited my chain, to mimic what was done there. Talk about a 3D stage! I only treated the mid component of mid/side processing. But the hard part is the tonality gets all messed up. Though with a lower inverted signal you can get some of it back to "normal".
What I did was create a center sum signal, invert it and band pass it and next combine it with the mid signal of my chain, about 17 dB below the original signal. It has the wanted effect of breaking loose the barriers of the sound, but it also alters perceived tonality. I immediately fired up REW to look at the signal sum at the ears, it still has cross talk dips, but the first wave hitting the ear now is an inverted image of that Dirac pulse. It really works though.
It filled my room out nicely, stage almost all around me basically. Now I've had that before, but this time the center played along. It is fun, but do I want to keep it? I may steel some ideas from it for a more toned down version.
Hard to say what route would be best, but I agree... a center channel would have been a smarter solution in the end.
If I get some free time I'll show some graph's of what's happening with the inverted phantom signal. It is fun as a learning exercise though. Prof. Choueiri's impulses finally make sense to me, even the wavy phase part of it. And it's easy to re-create in JRiver.
Got to do some more thinking I guess
.
Now that HD video has gotten cheap and easy, and surfing the web is starting to be available with many cable installations, it seems like it's just a matter of time before the typical stereo system has an HD video screen in the middle. At that point having a center speaker seems desirable. The only question in my mind is how to extract an L+R from the stereo feed such that there is reasonable apparent separation from the L and R speaker signals, without making the whole thing sound too mono.
My "Extractor" project is largely getting there... boards are loaded, chassis is fabricated and painted, I'm held up by dry transfer lettering that I ran out of, but that's now on order... Maybe it will be done in a month. These projects always take much longer to get done than I expected.
I do plan to share the results, and I do have some insanely picky tube guru friends that will give their worst possible criticisms (since they hate transistors, and my unit has 17 dual opamps (only four in the actual extractor circuit - the rest are in the "Holographic Generator" inter-aural cancellation circuit and the 30HZ boost circuit - each bypassable)). If they like the result, then maybe my Extractor is a success.
When doing critical listening at the sweet spot, I expect it to be quite good, and when I'm away from the sweet spot, I'll just want to hear everything well anywhere in the room, and I think it will do that well too (with the H-generator bypassed or BW limited). I've included 2 BW limiting options on the extraction processing (based on David Griesingers experiments) that may offer better fine tuning for a given set of variables. And the center output has a dedicated 4 section Baxandall tone circuit, since that speaker is very different than the L and R speakers in several ways.
My "Extractor" project is largely getting there... boards are loaded, chassis is fabricated and painted, I'm held up by dry transfer lettering that I ran out of, but that's now on order... Maybe it will be done in a month. These projects always take much longer to get done than I expected.
I do plan to share the results, and I do have some insanely picky tube guru friends that will give their worst possible criticisms (since they hate transistors, and my unit has 17 dual opamps (only four in the actual extractor circuit - the rest are in the "Holographic Generator" inter-aural cancellation circuit and the 30HZ boost circuit - each bypassable)). If they like the result, then maybe my Extractor is a success.
When doing critical listening at the sweet spot, I expect it to be quite good, and when I'm away from the sweet spot, I'll just want to hear everything well anywhere in the room, and I think it will do that well too (with the H-generator bypassed or BW limited). I've included 2 BW limiting options on the extraction processing (based on David Griesingers experiments) that may offer better fine tuning for a given set of variables. And the center output has a dedicated 4 section Baxandall tone circuit, since that speaker is very different than the L and R speakers in several ways.
Thanks to you, Bob, I'm actually on the road of Prof. Edgar Choueiri again. I wanted to understand his impulses from the time he was still involved in the Ambiophonics group effort. As I already apply a sort of cross talk cancelation to the sides, because of the mid/side EQ I apply, the missing link is the center part. After playing with it today I think it has potential.
What it does do is keep both left and right signals processed equally, as opposed to the shuffler that really creates a difference between the two, both in time and frequency.
So I added 2 cross talk chains, only to the phantom center. On paper it looks pretty good actually. I'm only aiming to fix the 1850 HZ dip and the 5500 Hz dip, as well as soften the bumps at 3700 and 7400 Hz. If what I simulated this evening is right it might be a better ticket than the shuffler. And I get to keep my time-coherency that way, still early though.
But as long as it's fun to find out, I'm going for it...
What it does do is keep both left and right signals processed equally, as opposed to the shuffler that really creates a difference between the two, both in time and frequency.
So I added 2 cross talk chains, only to the phantom center. On paper it looks pretty good actually. I'm only aiming to fix the 1850 HZ dip and the 5500 Hz dip, as well as soften the bumps at 3700 and 7400 Hz. If what I simulated this evening is right it might be a better ticket than the shuffler. And I get to keep my time-coherency that way, still early though.
But as long as it's fun to find out, I'm going for it...
I'm moving the opposite direction: I want the speakers to be at the same height as the screen image, so no room for a centre speaker...
After testing some variants of the cross talk cancelation and comparing it to the Two Bump Shuffler + a little mid-side EQ that last one still wins for me.
When you sit at one spot, don't move or turn your head, the cross talk can be very entertaining and more "3D" although it still is close. But the shuffler works better when you move off axis, no added coloration like the specific cross talk cancelation I used.
It's easier to be wowed for a couple of minutes than to keep enjoying something, that much I've learned.
While cross talk cancelation does work, I'm not willing to accept the coloration this simple algorithm brings. Too restricting in other words.
When you sit at one spot, don't move or turn your head, the cross talk can be very entertaining and more "3D" although it still is close. But the shuffler works better when you move off axis, no added coloration like the specific cross talk cancelation I used.
It's easier to be wowed for a couple of minutes than to keep enjoying something, that much I've learned.
While cross talk cancelation does work, I'm not willing to accept the coloration this simple algorithm brings. Too restricting in other words.
I'm still loving the line arrays for Home Theatre... an excellent combination. Though a couple of subs should relieve and support the bottom end.
yep ever since I first heard a set of line arrays (the "twenty five" design from K&T, eq'd by a Groundsound) I was sure I want them for my HT. I already have 2 dozen W4-1320's for that purpose (will not be full length) waiting to get prio....
With a few EQ tweaks I'm now getting the most 3D stage I've had to date.
The 2 bump shuffler is still in use for this. This beats everything I've tried so far to get a 3D stage. I mimicked the frequency result of the cross talk to get there. I have not tried it without the shuffler yet. It's nothing severe in EQ. Just a couple of bumps and peaks at the right frequencies. I tried a couple of songs, about 5, very different tracks and they all had a very enveloping sound with depth and width that I didn't experience before all at once. I've had parts of it on separate occasions, but not all at once to this level.
I still have to wait and see if this holds up over time. It may just be me.
The 2 bump shuffler is still in use for this. This beats everything I've tried so far to get a 3D stage. I mimicked the frequency result of the cross talk to get there. I have not tried it without the shuffler yet. It's nothing severe in EQ. Just a couple of bumps and peaks at the right frequencies. I tried a couple of songs, about 5, very different tracks and they all had a very enveloping sound with depth and width that I didn't experience before all at once. I've had parts of it on separate occasions, but not all at once to this level.
I still have to wait and see if this holds up over time. It may just be me
.
Just finished another thorough study of the phase-shuffler. This stuff is pretty tough to get right.
In other words, there were still little things that bugged me, despite the nice 3D stage and envelopment. No matter what I tried I could not get the tonal balance like I wanted it to be.
Using the same EQ without the shuffler lost me some of that 3D feel, but had a more pleasing tonal balance. It's cool though that I could find what I heard in my (prediction) plots.
The shuffler has one side that's actually working very well to eliminate the cross talk dips. But the other side has some countering results. No matter how hard I tried to find a solution to fix that, I'm not there yet. I tried various phase angles but could not solve this left/right imbalance to my liking.
I will try a few of the phase variants in real life testing when I get the time to play again.
The tool that seems to predict what I hear out in the room the best is the early waterfall plots that show the situation after 270 us has passed.
With one shuffler I noticed I had to re-adjust left/right balance... Those early waterfall plots showed me why. The phase and timing differences definitely have their influence. But it's not quite right yet.
In other words, there were still little things that bugged me, despite the nice 3D stage and envelopment. No matter what I tried I could not get the tonal balance like I wanted it to be.
Using the same EQ without the shuffler lost me some of that 3D feel, but had a more pleasing tonal balance. It's cool though that I could find what I heard in my (prediction) plots.
The shuffler has one side that's actually working very well to eliminate the cross talk dips. But the other side has some countering results. No matter how hard I tried to find a solution to fix that, I'm not there yet. I tried various phase angles but could not solve this left/right imbalance to my liking.
I will try a few of the phase variants in real life testing when I get the time to play again.
The tool that seems to predict what I hear out in the room the best is the early waterfall plots that show the situation after 270 us has passed.
With one shuffler I noticed I had to re-adjust left/right balance... Those early waterfall plots showed me why. The phase and timing differences definitely have their influence. But it's not quite right yet.
Interaural cancellation techniques cause the center image to be weak or "phasey". It just occurred to me why. I think it makes more clear the comb filtering our binaural hearing mechanism gets when trying to sum L and R acoustic signals from wide spaced L and R speakers. When there is crosstalk, it's analogous to many random room reflections filling in each others cancellations. When you reduce the number of acoustic sources (reflections or short interaural crosstalk delays), the remaining ones show their true colors more. So it seems that it's because we hear binaurally. I keep remembering that ancient Bozak speaker with the two vertical line arrays of tweeters... What would drive someone to doing that? Did it actually reduce the weak center image in the frequencies above about 1kHZ? Is it optimized for binaural hearing? Would that only work for the sweet spot?
Attachments
Last edited:
By the way, I'll post it here on my own thread and not make a fuzz about it on other threads...
Look at preferred full range speakers with claimed 3D imaging. One thing they will have in common is a (natural) bump around 1850 Hz. Even better if 3-4 KHz is slightly recessed. So far for the "magic" drivers... it's perfectly explainable.
All because we have 2 ears we listen with...
So don't make your curves perfectly flat, bump 1850 Hz 1,5 to 2 dB, dip at 3 to 4 kHz by the same amount, even better if you only do the mid part of mid/side processing.
(another higher Q dip at 7.4 kHz of about 1 to 1.5 dB is helpful too)
By the way, this also explains why the Vifa TC9 FD18-08 I use, you know, the cheap TV speaker, has a reputation of being flat and lifeless.
It has it's bumps and dips at the above numbers reversed. Perfectly curable with EQ to make it an excellent performer with an above average clean impulse. Making it very suitable for the line arrays.
Look at preferred full range speakers with claimed 3D imaging. One thing they will have in common is a (natural) bump around 1850 Hz. Even better if 3-4 KHz is slightly recessed. So far for the "magic" drivers... it's perfectly explainable.
All because we have 2 ears we listen with...
So don't make your curves perfectly flat, bump 1850 Hz 1,5 to 2 dB, dip at 3 to 4 kHz by the same amount, even better if you only do the mid part of mid/side processing.
(another higher Q dip at 7.4 kHz of about 1 to 1.5 dB is helpful too)
By the way, this also explains why the Vifa TC9 FD18-08 I use, you know, the cheap TV speaker, has a reputation of being flat and lifeless.
It has it's bumps and dips at the above numbers reversed. Perfectly curable with EQ to make it an excellent performer with an above average clean impulse. Making it very suitable for the line arrays.
Last edited:
I sort of gave up on the shuffler. Not that it doesn't do anything positive, just that it adds another compromise. I've reverted back to mid/side EQ, which does work a bit better for me, personally because it keeps tonal balance in tact in a better way than all of my other experiments.
Yet the shuffler did show me a positive change in perception I have not been successful to completely achieve with EQ alone. In my many nights trying to figure this out it points to the shuffler making a positive change on one side to combat the comb filtering. Though as the other side is reversed timing wise the problem on that side gets bigger. Meaning the response at one ear is different from the other one. One is relatively smooth in the first wave front, the other one has comb filter dips. This mixed signal does truly sound different, better in some ways (imaging) but worse in others (tonally).
The EQ solution does better tonally but is a bit different in perception. For me tonal balance is of a higher priority than imaging (which still is very good, just a bit different in the center compared to the sides). The shuffler had/gave a better sense of depth, hard to explain.
The search continues
Yet the shuffler did show me a positive change in perception I have not been successful to completely achieve with EQ alone. In my many nights trying to figure this out it points to the shuffler making a positive change on one side to combat the comb filtering. Though as the other side is reversed timing wise the problem on that side gets bigger. Meaning the response at one ear is different from the other one. One is relatively smooth in the first wave front, the other one has comb filter dips. This mixed signal does truly sound different, better in some ways (imaging) but worse in others (tonally).
The EQ solution does better tonally but is a bit different in perception. For me tonal balance is of a higher priority than imaging (which still is very good, just a bit different in the center compared to the sides). The shuffler had/gave a better sense of depth, hard to explain.
The search continues
Last edited:
It seems like a shuffler would damage the accuracy of in-between imaging below 1kHZ, where we judge image location more by phase or timing comparison than by amplitude comparisons in the upper midrange.
A little off subject but if I didn't already mention it, my audio hobby website is back up on the web. You might be interested in reading about my 2 > 3 channel "Extractor" project, which isn't quite done yet, but is getting there. Or the Deuce Jr. Stereo Guitar Amp.
Bob's Website
A little off subject but if I didn't already mention it, my audio hobby website is back up on the web. You might be interested in reading about my 2 > 3 channel "Extractor" project, which isn't quite done yet, but is getting there. Or the Deuce Jr. Stereo Guitar Amp.
Bob's Website
The shuffler isn't shuffling below 1 KHz, the biggest phase difference is a distance in cm of about 3.5 cm as explained earlier, I think, and listening to cross talk cancelation confirms that, the shuffler does act as a crude cross talk cancelation, but only effective on one side. I say crude, because the timing actually varies.
I can listen to it all day with eyes open. Yet it does present strain with eyes closed. As if my brain is trying too hard to make sense of the mixed queues. On the contrary, a cross talk cancelation, only applied to my phantom mid signal made me more at ease. Though it had it's own share of tonal issues. The paper by Prof Edgar Choueiri does present an explanation for that, though to solve it takes a lot of manipulation and would work best with personified HRTF measurements. It does make a lot of sense though.
I need not worry about any of this. These speakers image like champs, sound awesome at the sweet spot, have a very nice tonal balance and make speakers and room disappear. Leaving you with nothing but a very nice sound scape. It's more like listening to sonic events than playing a song. I have nothing to complain about.
It's just way to much fun to see if improvement is (still) possible. I have build myself the ultimate toy to experiment with. Learned a lot in the whole process. Had many chills and even emotional connections with many songs and genres. Every time a song hits me straight in the heart I know why I love doing these experiments. And believe me I've had my share of those moments. I often get lost completely in listening and simply enjoying.
But the one thing that these experiments do get me is training in listening, understanding my measurements and why these differ from what I hear and an education to boot.
I'll look at your website another time, it's way past bedtime (again). Hope you crack the code.
I can listen to it all day with eyes open. Yet it does present strain with eyes closed. As if my brain is trying too hard to make sense of the mixed queues. On the contrary, a cross talk cancelation, only applied to my phantom mid signal made me more at ease. Though it had it's own share of tonal issues. The paper by Prof Edgar Choueiri does present an explanation for that, though to solve it takes a lot of manipulation and would work best with personified HRTF measurements. It does make a lot of sense though.
I need not worry about any of this. These speakers image like champs, sound awesome at the sweet spot, have a very nice tonal balance and make speakers and room disappear. Leaving you with nothing but a very nice sound scape. It's more like listening to sonic events than playing a song. I have nothing to complain about.
It's just way to much fun to see if improvement is (still) possible. I have build myself the ultimate toy to experiment with. Learned a lot in the whole process. Had many chills and even emotional connections with many songs and genres. Every time a song hits me straight in the heart I know why I love doing these experiments. And believe me I've had my share of those moments. I often get lost completely in listening and simply enjoying.
But the one thing that these experiments do get me is training in listening, understanding my measurements and why these differ from what I hear and an education to boot.
I'll look at your website another time, it's way past bedtime (again). Hope you crack the code
.
Thanks a bunch. Being able to "chew the fat" and share by projects with competent audio engineers from all over the planet is a huge thrill for me. When I worked at Tektronix, Dolby Labs, and a few AV companies, there were always people to run ideas past and get criticisms from. In recent years (semi-retirement) it's become few and far between, so this website is much appreciated. I just wish I knew how to make a living doing this stuff... so much of this industry has left the U.S. What's left is competing with the very cheap Chinese labor and apparent lack of environmental laws.You deserve a friendly