I gave a list of songs to check out that are mainstream artists done in binaural. If you sue a streaming service you may try some out. Binaural isn’t a must to hear the speaker. It is just the kind of recording that an XTC speaker works best with. It would have the most extreme imaging.
the head shadow is actually the shadow case by the protrusion of the head. It’s the effect on the frequency response of a soundwave that hits the right ear for sound coming from the left side (or reverse). Since soundwaves get longer at lower frequencies and will increase in their likelihood of wrapping around the head, we have a rollled off response like that of a first order filter. But it isn’t exactly a first order filter. What Polk did was take advantage of the off-axis response of the array drivers by placing them at an extreme angle to the listener.
View attachment 46534
Here you can see the measured head shadow response at different angles.
as for the hafler circuit discussion. Are you now in agreement that this is being used to cancel crosstalk or are you suggesting something else? I notice you are conspicuously not mentioning that fact and I want to be sure we are in agreement here. The circuit isn’t meant to extract ambience information and play that back. It’s meant to cancel the crosstalk.
patents don’t need proof and I would never expect it. That is for white papers and journal articles. They haven’t published that. Which as much as we prefer to see that, is incredibly uncommon. I’ve been told recently that even some of the most prolific CE companies for publishing research will soon stop in favor of proprietary research.
for proof of the idea I would look at the many articles published on crosstalk cancelation generally.
In terms of how to measure such a speaker. It really isn’t that hard if you have the right tools, software, and facilities. The main response of the speaker matters most so you treat the speaker like any speaker. You measure it with the XTC function disconnected and measure it’s full polar response. I did that and presented the data.
crosstalk can be measured a number of ways. First is to measure IACC using a binaural measurement system. You can also measure more simply the ILD. The BAACH folks have further refined this into a direct measure of XTC in their software. There are a number of challenges for measuring XTC in a room however which is why I didn’t publish the results. Scott looked at them and questioned how confident we could be that they were right. They too had trouble getting good results and they had an anechoic chamber, which is a much better way of testing the effect of the circuit. We decided that since people wouldn’t be used to seeing this information and because there was a chance it wasn’t fully accurate, we wouldn’t publish that data. I am hopeful to use a more advanced piece of software from the folks at BAACH to measure it in the future. The software I used came from Farina and wasn’t really intended for this purpose. I also don’t have a HATS and was relying on crude gear that was showing its limits in these tests. My dummy head didn’t have a good head shadow and my ear mics didn’t have a flat enough response, weren’t matched, etc. I now have a much better head worn mic for this.
as for how to measure the response at the ear. That is it’s own science but for folks who do that, the response is well known. Typically you wouldn’t do that to measure a speaker. In this case it’s just to see how effective the crosstalk cancelation is and to see how significant the coloration actually is at the ear. How it would show up at an omni mic in room is not necessarily representative.
As for the room messing things up. Yes it does but it’s not as bad as you would think. You can still realize in excess of 15dB of XTC over a wide bandwidth with many of the systems on the market in a real untreated room. BAACH achieved in excess of 30dB and even this system appeared to be achieving just under 20 for me. The key is that the reflections need to be relatively weak and farther out in time. So a sufficiently large room or well absorbed lateral reflections. In my case, a single 4” panel fixed the problems in my room. As I understand, Polk doesn’t treat their rooms. Their demo rooms were simply larger than mine.
as for the narrowness of the sweet spot. James and I tested this and while it’s true that it’s a problem, it too may not be as big a problem as you might think. The farther you sit back the farther you can spread the speakers and the wider the sweet spot. So in a large enough room, you could achieve a sweet spot that encompasses multiple seats. In my theater you could but it wasn’t terrible either. The effect was not destroyed to the immediate side and we found if you moved back a row it sounded better. It also didn’t sound bad to the sides. It simply lost the amazing imaging and sounded like a typical speaker.
while you are right that modern upmixers can do a lot. I don’t see this as an either or. This system can work in multichannel. As another noted, this is a core part of modern ambisonics. XTC in surround upmixers is not a new idea for the consumer market either. But I could see this becoming more core as technology progresses. The advantage would be a larger sweet spot and far more immersive experience. These systems rely on precisely placed and characterized speakers. They work best with highly controlled and fairly narrow dispersion speakers so that is probably the main reason we don’t see them being marketed as a legit technology. Outside of a lab you can’t control what speakers people buy or how they set them up. But as I said. I’ve heard a full ambisonics system and it was simply the most realistic recreation of a musical performance I ever heard.
I think I need to go back to first principles again. This approach is bogus. I have thought about this long and hard.
After looking at the circuit, I accept that because of the circuit based on the Hafler principle that signals of equal intensity will not be cancelled.
However what I do not accept is the principle in right left crosstalk has to exist. That fact that it is an issue, is because recording engineers by and large do not know their craft.
So let's go back to first principles. In a purely two channel system, the location of any sound can be defined mathematically by x/y geometric coordinates. If it is a 3D system like Atmos then it is defined by x/y/h coordinates. This is the whole foundation of the Atmos system and why it works.
Now I would maintain that the human brain deals with crosstalk continuously. People with normal hearing hear all sounds with both ears.
Now in a stereo recording done correctly, any sound can have the x/y coordinates preserved and recorded correctly. This requires that the relative intensities of sounds as to their left and right coordinates be preserved correctly. So if a sound is central, then speakers will play equally. If the sound is between the relative intensities will be preserved to the degree they differ from the 90 degree datum to left or right, and the mathematical coordinates will be preserved.
Now this requires, that these relative intensities be recorded at a point in space.
Alan Blumlein knew this back in 1935. We continue to disregard the inventor of stereo sound, and one of the pivotal geniuses in the history of audio.
His point was that if you record at a point the relative intensities of sounds relative to position then you preserve the coordinates of the origin of the sound. This assumes they you do not cause gross time shifts throughout the recording chain.
However what to we do? We place a bunch of microphones all over the sound field widely separated in space, and played back via speakers generally plagued with unwarranted time aberrations. Then there should be absolutely no surprise imaging is vague to non existent.
So Alan Blumlein reasoned correctly that what was required for good imaging was a microphone system that recorded the intensity of sounds relative to their left right position. His reasoning can not be faulted, yet we continue to ignore it. Hence we go in search of nonsensical solutions to a problem of our creation.
I can assure you that if you follow Blumlein's logic the problem is solved.
I can demonstrate this, and will post here again, some of my Blumlein intensity stereo recordings I made over the years I made radio broadcasts. These image perfectly. They were actually really pivotal in guiding my speaker design development. Actually there really do sort out speakers. If a set of speakers can present an accurately localized images between the speakers with some depth of field, with an even FR, they are winners. If they can't, they are defective in design, no ifs or buts.
Now with the added bonus of the new Dolby Digital surround upmixer, these recordings produce an incredibly lifelike representation of the original venue on this 7.2.4 system. I have no need of center spread. In fact it ruins the effect. If anything left right imaging is actually enhanced.
So I will post again these recordings about which I also give background information. I would be interested to see how they sound on those Polk speakers. I have a feeling it may not be pretty. You should try them with an without the outside speakers engaged. When used as a straight stereo pair, they should be 12' apart though and not 6'.
These are the two videos I put up and you can see the metering, including the phase scope.
I'm also going to post another of Tom Scotts Blumlein recording of his brother playing Wachet Auf, on the organ at Petworth Sussex.
This shows how well this mic technique preserves spatial information. This organ has a division on the right and a division right ahead, as you can see in the video.
The flutes are played in the right division, and the reed playing the melody is right ahead. Even though this is a two channel recording, the flute division is firmly in the right of the room, and the reed up front, with very little crosstalk. So the right surround is totally dominant in the reproduction of the right flute division. At least it is on this rig.
