I became frustrated with that, and looking for something to show colourations, wrote a raytracer to use measured data. As interesting as it was, I can't say I came up much the wiser for it though.
Earl indicated once that raytracing below FSchroeder may be less than accurate so I mostly rely on the sub setup procedure to get that band right, and I wouldn't listen to a colouration calculator for higher frequencies.
In the midrange there is a conflict between colourations produced with and without significant group delay so there is going to be a need for some room treatments there anyway.
Blauert writes about trading in "Spatial Hearing". The effect depends on signal type and on signal level.
Yes, and let's not forget the variation between the persons.
It propably could be made to give some effect in special circumstances and for some persons, but it is not a robust method to give general improvements.
- Elias
Unfortunately, the term "time intensity trading" is being mis-used in this thread. In the research community (Blauert and others) it generally (but not always) refers to something a bit different.
I think we use it correctly. I'm not talking about equivalence but about trading, i.e. how much louder does the left or right speaker need to become to counterbalance the change in arrival time and level when moving the listening position to the left or the right.
There's a nice tool by Sebastian Merchel which processes the L and R signal in real time based on listener location. It gets that location from a front mounted camera.
Please see SweetSpotter
Unfortunately I couldn't get it to run.
The concept of time-intensity trading via toe-in has been published as early as the mid-1950s. Attached is a link to Ted Jordan's article of 1971 in Wireless World, but I recall reading a similar article in WW from the 1957 time frame. I THINK it was also by Jordan, but not positive. I did get the impression that the mid-1950's article was repeating what was already known.
I have never heard a stereo pair that did not benefit from toe-in as a technique to broaden the sweet spot and/or help imaging. Perhaps a CD polar pattern would give the widest sweet spot, but it works for any and every speaker pair I've used the concept on to get the widest sweet spot, and the best imaging, that can be had with that particular pair. And NO, it does NOT create a headphone like in-the-head image. Don't believe anyone who tells you that!
It is known that some people can't perceive a correct phantom image. So if someone tells you this technique doesn't work, take it with a grain of salt. It works very well. It is one of the first order setup criterias I use for loudspeakers. I don't know how a computer program would do any better than actual listening. With so many variables between various loudspeakers, especially varying polar patterns between drivers and the effect of crossovers, it would seem an impossible task. Not pointless, if a general understanding of what is happening for physical reasons. But ultimately, it's about listening to the pair being toed in. It may be a classic example of listening as opposed to measurement.
http://www.ejjordan.co.uk/PDFs/Jordan_WW_Feb_71.pdf
I have never heard a stereo pair that did not benefit from toe-in as a technique to broaden the sweet spot and/or help imaging. Perhaps a CD polar pattern would give the widest sweet spot, but it works for any and every speaker pair I've used the concept on to get the widest sweet spot, and the best imaging, that can be had with that particular pair. And NO, it does NOT create a headphone like in-the-head image. Don't believe anyone who tells you that!
It is known that some people can't perceive a correct phantom image. So if someone tells you this technique doesn't work, take it with a grain of salt. It works very well. It is one of the first order setup criterias I use for loudspeakers. I don't know how a computer program would do any better than actual listening. With so many variables between various loudspeakers, especially varying polar patterns between drivers and the effect of crossovers, it would seem an impossible task. Not pointless, if a general understanding of what is happening for physical reasons. But ultimately, it's about listening to the pair being toed in. It may be a classic example of listening as opposed to measurement.
http://www.ejjordan.co.uk/PDFs/Jordan_WW_Feb_71.pdf
pooge, I can perceive phantom images just fine. Toe-in still doesn't work for me. Guess we have to take your comment with a grain of salt too
So exactly what is it that doesn't work for you?
A phantom center jumps to the nearest speaker when I move to the left and the right just like it would with the standard 60° triangle.
It is more stable with a matrix decoded center. But even then the sweetspot is rather small. Guess my hearing is very sensitive to ITD.
I've animated the pre-after toe-in pictures provided by Earl. The red point is the 60° stereo triangle sweetspot:
You see how it works better, the more you go behind the 60° stereo triangle
Rudolf
The phantom will still move, and still favor nearest speaker if you move far enough to left or right. What the toe-in does is minimize this effect as much as possible, and thereby increase the sweet spot as much as possible and lock in imaging for a given pair. It isn't done in place of the triangle. It is done WITHIN a proper equilateral triangle to create an optimum sound field for imaging a Blumlein recording.
It may take a bit of experimenting to get right for a given pair, but once right, you know it. Start by aiming the right loudspeaker to the left most seating position, and the left speaker toward the right most seating position. If you have a wide seating area, this might not be optimum for imaging and/or frequency response in the center position. Play a mono recording of a single voice, or one panned to the exact middle. I found Gordon Holt's recording of "Why HiFi Experts Disagree" on a Stereophile test CD to be useful because it goes on long enough to get results. Sit in the middle seat. Move your head around by turning your head side to side. If the image shifts with that little movement, reduce toe-in until the image is stable with that little bit of movement. Then lean your body side to side, and adjust toe-in so these larger movements are stable. You may not get stable images past this amount of movement away, depending on your particular speakers, but it will probably still be the best toe-in for far off center positions that you speakers can do, and still have the best and stable image from the center seat.
I find a single voice is best if the recorded voice didn't move while recording, as in someone walking across the stage in an opera. When you listen to someone sitting in one place, like Holt's recording, you don't EXPECT them to be moving wildly around with head position. In fact, it's a rather disconcerting reminder that you are listening to a pair of loudspeakers rather than the real thing. Whereas, with a recording with a full orchestra, that image shift with head movement doesn't stand out as much. Yet, the orchestras do sound better with proper toe-in. It's just much harder to use as a tool.
Get the toe-in correct, and you'll know it. The imaging can be holographic!
It may take a bit of experimenting to get right for a given pair, but once right, you know it. Start by aiming the right loudspeaker to the left most seating position, and the left speaker toward the right most seating position. If you have a wide seating area, this might not be optimum for imaging and/or frequency response in the center position. Play a mono recording of a single voice, or one panned to the exact middle. I found Gordon Holt's recording of "Why HiFi Experts Disagree" on a Stereophile test CD to be useful because it goes on long enough to get results. Sit in the middle seat. Move your head around by turning your head side to side. If the image shifts with that little movement, reduce toe-in until the image is stable with that little bit of movement. Then lean your body side to side, and adjust toe-in so these larger movements are stable. You may not get stable images past this amount of movement away, depending on your particular speakers, but it will probably still be the best toe-in for far off center positions that you speakers can do, and still have the best and stable image from the center seat.
I find a single voice is best if the recorded voice didn't move while recording, as in someone walking across the stage in an opera. When you listen to someone sitting in one place, like Holt's recording, you don't EXPECT them to be moving wildly around with head position. In fact, it's a rather disconcerting reminder that you are listening to a pair of loudspeakers rather than the real thing. Whereas, with a recording with a full orchestra, that image shift with head movement doesn't stand out as much. Yet, the orchestras do sound better with proper toe-in. It's just much harder to use as a tool.
Get the toe-in correct, and you'll know it. The imaging can be holographic!
I've animated the pre-after toe-in pictures provided by Earl. The red point is the 60° stereo triangle sweetspot:
View attachment 335156
You see how it works better, the more you go behind the 60° stereo triangle
Rudolf
Semantically, I prefer to state it as working better with more toe-in rather than going behind the 60° stereo triangle. The image width will collapse if you go behind the triangle. You don't want that. You want to stay around the point of the triangle to get proper width of image.
The points I have been making are more geared to best sound from a center seat. Home theater needs for many seated people might alter these criteria for best compromise of intended use, depending on speakers used.
I said "You see how it works better ...". That's semantically as precise as it gets. Because I don't talk about the acoustical effect, but about the visual impression that one gets from the two pictures.
If it doesn't correctly represent, what happens to the ear, Earl probably needs to correct the graphical implementation.
If I were Earl, I would change the color scale to go from black over dark lilac to red, orange, yellow and white. This should allow to give a consistent impression of going from dark to light too.
Rudolf
Markus - this is correct and we are using the concept correctly. But in Blauert he uses a concept of degrees which I find hard to equate to the fures that you show unless those figures are for detection. But a complete pull to the closest speaker with less than 1 ms of delay diferential would not agree with Blauert at all. This would be in the very small deviation region. Thats why I am trying to understand the figures that you posted.
If there is a better model than Blauert then I will use that one, but in my searching I have not found one except a very complex model that would be overkill for what I am trying to do. I read a recent paper from one of Blauerts students on this subject and in that paper they use only the ITD and ignore the ILD. The ITD is the stronger factor, as my equation shows, but ignoring the ILD would not allow for "trading" at all.
I also noted a paper that derived the "optimal" directivity pattern. Then they created a DSP system for a pair of speakers to impliment this directivity. They showed a comparison of the predicted directivity and the measured directivity which showed the two to be completly different. This is exactly why I am not so much interested in the "ideal" directivity as much as how effective can two "real" loudspeakers be made to work.
I've animated the pre-after toe-in pictures provided by Earl. The red point is the 60° stereo triangle sweetspot:
You see how it works better, the more you go behind the 60° stereo triangle
Rudolf
Thanks Rudolf
In my software this change happens in real time for both the angle of the speakers and the seperation. There is an optimum, but your animation makes the point almost as well. At first I was surprised by how much the situation improves when the speakers are moved closer together. But then I remembered how strongly the effect depends on delay differences and it made perfect sense.
To avoid "the impossible task", I use measured data of the real loudspeaker taking this complexity out of the problem. Then a "general understanding" of what kinds of speakers work and how best to set them up is gained from the software. One could just buy all the speakers on my list and try them by listening, but that seems like a rather arduous task. The software is a lot easier.
I have experimented with the forward and back positions and I agree that one does not want to be too far back of the point of the triangle. The stage width minimizes even if you are still in the sweetspot.
That would be the algorithm not the graphics. I admit that there may be better algorithms, I just havn't found them.If it doesn't correctly represent, what happens to the ear, Earl probably needs to correct the graphical implementation.
I read a very in depth paper on the use of color in graphics like this. The colors were precisely choosen to go from brighterst to darkest in a smooth and continuos manner and they do that. Now should "good image" be bright or dark? That is basically the only option available.
I remember that you put some emphasis on that - and rightfully so.
It's like real life: The good ones are always bright and the bad ones always dark.
I believe the culprit is the diagonal shading of the blocks. To me it makes some "outer" blocks look brighter than the adjacent "inner" block. But that is just cosmetics.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.