This graph looks like the measurements were taken with the microphones in ear, those erratic spikes and dips look like pinna effects.
The graph seems to indicate the difference in those properties rather than the HRTF.
It shows a downward slope of the red and purple signals, but I believe the graph is pointing out the inconsistency of left and right pinna shapes.
What site were these graphs taken from?
Last edited:
HRTFs are caused by properties of the head, torso, pinna and ear canal so I don't understand your comment.
Ok, sorry. I'm talking about the head shadow.
Because there is nothing we can do about the difference in the shape of our ears.
That's how we are trained to localize things in real life.
He was speaking of ILD, so that is mainly the function of the head shadow.
Which is funny because the graph looks just fine where ILD is dominant.
Oh, I edited my post because the microphone couldn't have been placed in the canal because there would have been a huge spike around 3 kHz.
Because there is nothing we can do about the difference in the shape of our ears.
That's how we are trained to localize things in real life.
He was speaking of ILD, so that is mainly the function of the head shadow.
Which is funny because the graph looks just fine where ILD is dominant.
Oh, I edited my post because the microphone couldn't have been placed in the canal because there would have been a huge spike around 3 kHz.
Last edited:
Plots from:
Single Bipolar Loudspeaker System for Stereo Reproduction
Fredrik Gunnarsson
CTO Embracing Sound Experience AB
I think you are basically on the right track, however..
The graph is not really representive of a stereo-dipole situation.
There seems to be too much attenuation at 100Hz probably as a result of near-field effects.
Any asymmetry btween ears is not an issue here, since it is the same ear for both the main and cancelling paths.
It is a matter of the magnitude of the change in 'gain' for small directional changes due to the pinnae, compared to the mean head shadow effect, and that can not be implied from a graph at +/- 30 deg., near field.
I think it is an issue, despite my original assumption that the sound was tangential to the head for both paths.
The original ISVR, as I understand it, did use HRTF data, but although impressive, was ultimately unlistenable for audiophile ears.
I have certainly played with small barriers, which seemed to have more effect on the bass. Fine for DIY, but is not going to fly for the public at large.
One problem with using HRTF data is that it is only going to be useful if the tweeter has near perfect response. Another is that if you move your head, it is probably more unpleasant to have the sound change, than to have it biassed towards the front, but less sensitive to movement.
If this graph is showing the difference in pinna's on an individual, then it is not referring to path lengths to each ear or the head shadow.
These two factors are what are responsible for ILD and ITD.
It is showing the phase inconsistencies of the folds of individual pinna's.
The microphone of course pics these up as having very different frequency responses.
The point is that our brains interpret the differences as sounding the same.
So with this in mind, electronic xtc should work just fine.
If there is a particular sound emanating from the left speaker, and is being cancelled with the same sound in opposite phase on the right speaker, then our brains interpret these as identical sounds even though they look very different on the graph that the mic's are displaying.
I don't mean to argue, just to understand.
These two factors are what are responsible for ILD and ITD.
It is showing the phase inconsistencies of the folds of individual pinna's.
The microphone of course pics these up as having very different frequency responses.
The point is that our brains interpret the differences as sounding the same.
So with this in mind, electronic xtc should work just fine.
If there is a particular sound emanating from the left speaker, and is being cancelled with the same sound in opposite phase on the right speaker, then our brains interpret these as identical sounds even though they look very different on the graph that the mic's are displaying.
I don't mean to argue, just to understand.
I think we are making this too complicated.
Phase effects amplitude at the microphone.
I believe that is why the response is different in pressure level at different frequencies.
This experiment should have been done excluding the pinna.
Only the head shadow to show differences in ILD and ITD.
Then we would see that they are pretty constant.
Phase effects amplitude at the microphone.
I believe that is why the response is different in pressure level at different frequencies.
This experiment should have been done excluding the pinna.
Only the head shadow to show differences in ILD and ITD.
Then we would see that they are pretty constant.
I think you will find it is easier to understand XTC in terms of impulses. You simply have to add one impulse to the other (with regard to sign)
If you download the race impulses from my filter diectory and look at them in an audio editor you can see what happens, with a bit of thought. In order to cancel the two channels from left and right speaker, they need to end up as a mirror image of each other, so they add to zero for the far ear. By attenuating one by 2dB and shifting 2 samples this happens. Do it the other way round you end up with a single-sample pulse ie. a flat frequency response for the near ear. (assuming no pinna effects)
Ok, well that's why I said that the measurements in the graph should have been taken without the pinna.
I understand xtc.
Those irregularities would show up with a barrier, with normal stereophonics, ambisonics, or any other method of sound reproduction, because they are irregularities of the pinna. Not the head shadow, which is the cause of ILD and ITD.
I'm very perplexed by this conundrum!
I may change my position the more I think about it. I don't want to be steadfast on any decision about this.
I need a few days to wrap my head around this malarkey.
But who knows, I could have been right!
Haha
I may change my position the more I think about it. I don't want to be steadfast on any decision about this.
I need a few days to wrap my head around this malarkey.
But who knows, I could have been right!
Haha
The pig in the oinkment is that the head is not really a 2dB linear-phase attenuator.
I think here is the biggest hurdle of any cross talk cancelling system - high freqs.
Obviously using very accourate high freq HRTF data works only for fixed head positions, but the head turning (subconsciously) collapses the performance.
And if not using HRTF, the high freq sounds come from the central location.
In this sense it is not much different from a conventional stereo triangle ! In stereo triangle the high freqs are mostly perceived as coming from the speakers if direct sound dominates. This is my experience. The same for a stereo dipole. However, the advantage of a stereo dipole over stereo triangle is most of the sounds are mixed in the center anyway, so error is statistically less occuring.
There is another kind of a solution. Simply make the high freqs diffusive ! So no direction is preferred over the another. Then high freqs will not generate any conflicting directional cues, but imaging is based on midrange freqs using the available cues. I have employed this concept quite succesfully with stereo speakers. The key is to use the room reflections as a benefit. One must relax the fanatism on requirement of direct high freq sound ! But how, because usually stated is the Ambiophonics requires very damped room !?
This high freqs problem turns out to be a loudspeaker design issue rather than a cross talk canceller issue ! Why these two domains, speakers and cancellers, are treated like a separate entities ? I dont understand.. One day loudspeaker designers and cross talk canceller designers should shake hands and join the forces !
- Elias
Elias;2911862 There is another kind of a solution. [B said:
There are ways..
You may have missed the point that for a fully XTCing filter, The relationship between the direct sound at the ears, and the speaker feed levels, (and thus at wavelengths << than the speaker separation, the diffuse sound ) is entirely determined by the chosen delay/speaker-geometry and head geometry. Chosen carefully, you can put a notch in the direct sound relative to the reflected sound, which would help.
My room is ordinary but I place the speakers at about 1 metre or closer. Is is important to get the right balance between direct and reflected sound. I can get distance back with my Knobs(tm).
Like "all channel stereo" sounds so much better than normal stereo?
I'm not convinced that adding even more conflicting cues will improve overall sound quality.
I'm aware that with cross talk cancelling and HRTF manipulation one can create any imaginable spatial perception. However it requires accourate individual HRTF, and precise knowledge of the head position. Neither of these are not generally available for people listening at home environment. Thus I think the system should not be based on these assumptions. Then it automatically follows that electronic cross talk cancellatation do not work very well for high freqs (because no HRTF), and head turning will degrade the performance (because no head position information, though the barrier works better).
Is this the reason why Ambiophonics is not more popular ?
You have special listening habits 🙂 Personally I dont like to stare any nearby objects in front if me 😀 Usually I listen at 3-5 metres, and I do require a large listening area (Not a spot!). Perhaps for these reasons I had to abandon Ambiophonics many years ago already
I do keep following the development however because I find it technically interesting and psychoacoustically one can always learn something new.
- Elias
Like "all channel stereo" sounds so much better than normal stereo?
I don't consider diffuse field as a conflict. Conflict is where you got two or more clearly perceived spatial cues for a same sound event but they are pointing to different directions, for example.
- Elias
Seems to be there are other commercial speakers employing Embracing technology.
Geneva Sound Systems
There is cross talk cancellation if sitting in front of the speaker.
- Elias
Geneva Sound Systems
An externally hosted image should be here but it was not working when we last tested it.
There is cross talk cancellation if sitting in front of the speaker.
- Elias
Here again the conceptual nearfield radiation pattern.
Basically it's a MS (mid side) stereo loudspeaker, but cross talk is cancelled at the near field center line.
In the pic the signals are:
Dark blue = L + R
Light blue = L - R
Yellow = R - L
When these sum at the listening position, left ear gets L and right ear gets R stereo signals, thus cross talk is cancelled.
- Elias
Basically it's a MS (mid side) stereo loudspeaker, but cross talk is cancelled at the near field center line.
In the pic the signals are:
Dark blue = L + R
Light blue = L - R
Yellow = R - L
When these sum at the listening position, left ear gets L and right ear gets R stereo signals, thus cross talk is cancelled.
- Elias
Last edited: