Hello,
the principle of the wave field synthesis until today is implemented by loudspeaker rows in ring around the listener. The rendition remains two- dimensional therefore. I gladly would discuss my idea for the physical reconstruction of the sound field in all three space dimensions by a speaker wall in front of the listener. The procedure is described by http://www.syntheticwave.de (http://www.syntheticwave.de) . The first site shows an animated and easily understandable description of the WFS principle.
Regards Helmut
That was a very interesting read. The English version has some translation problems at times and can be a bit hard to follow, but is generally fairly clear. A couple of questions:
- The procedures you describe are fundamentally very similar to active beamforming as applied to antenna arrays for radar and communications applications. Several excellent beamforming algorithms exist for 2D antenna arrays, and I can't see how they wouldn't work for sound. Have you included antenna beamshaping in your literature search?
- Because of the aperture of your array elements (fairly large) and the size of the average listening room (fairly small), you are doing your beamforming in the "near field" of the radiators. It is my understanding that near field beamforming requires a precise understanding of the phase response of the radiators, instead of just a knowledge of the position of the phase centers. I have no idea if this is also true with accoustic sources, but is it something you have considered? Bear in mind that the vast majority of speakers behave very differently from an ideal dipole.
- Are you sure that there will always be a solution to the pattern synthesis problem, including wall reflections? I have never thought about the problem before, but the addition of multipath effects will add a more new constraints to the problem without increasing the number of degrees of freedom. It is possible that for a given synthetic source at a given position with a given bandwidth, these constraints prevent a solution to the beamforming problem from being found (or stop the optimiser from converging on an approximate solution).
Beamforming and pattern synthesis are not my field, so I might be wrong about some of these points, so don't take them too seriously.
Do you have plans for formal publication of your results in a peer reviewed journal?
That was a very interesting read. The English version has some translation problems at times and can be a bit hard to follow, but is generally fairly clear.
Hello cabbagerat,
I see you and several other hydrodgenaudio users very extensive visit the www.syntheticwave.de site, thank all for interest and excuse me for some mistakes in english. We old Germans have only education in russian.
The topic indeed meets the antenna array beamforming technologies, except the propagation speed of the waves. I am sure, some of the principles would be applicable for audio beamforming. Also acoustic researchers work on that topic successfully. Especially the IEM Gratz in austria can provide highly interesting results yet. I think also for radar you must provide the same amount of wavelengths inside the antenna group for all frequencies for balanced directive characteristics. That is also the principle of 1Ltd. Anthony Hooleys sound projector. For a enlarged field would decline the problem by my opinion. Its not very important, if 56 or 99 wavelengths fits into the field, the problems would arise below 300...400 Hz. In that range some unsolved problems exist, but I see the core range for directive cues in the 300..3500 Hz range. Above we currently indeed should stop the synthesis procedure and let due two ribbons its work.
The phase shifting of the speakers would generate no new problems, as far as equal for all speakers. The same time delay occur by same frequencies.
What distinguish the radar from audio is, we can use psychoacoustic rules in addition. As far as we can avoid misguiding cues before the direct wave hits the listener, the success is established by far.
We have a very high level to take a notice of faults in audio range. And the result of the transformaitin procedure should much more hit the genuine relations.
The main opportunity by the transformation procedure I see in the fact, not to finish the transmission chain by the loudspeakers, but by listeners ear.....
Regards in the sunny South Africa from cold Germany
Helmut
I see you and several other hydrodgenaudio users very extensive visit the www.syntheticwave.de site, thank all for interest and excuse me for some mistakes in english. We old Germans have only education in russian.
No problem. Your English is much better than my German or Russian
Its not very important, if 56 or 99 wavelengths fits into the field, the problems would arise below 300...400 Hz. In that range some unsolved problems exist, but I see the core range for directive cues in the 300..3500 Hz range. Above we currently indeed should stop the synthesis procedure and let due two ribbons its work.
Of course. The difficulty of the beamforming problem increases with the percentage bandwidth of the signal (or the number of octaves). Even handling 300Hz to 3.5kHz is nearly four octaves, so the challenge is large. How much research has been done looking at the frequency bands which are key for direction perception?
Obviously the low frequencies aren't and the extreme high frequencies can't be (the phase shift is too much), so I would guess that frequencies where the wave length is on a similar order to the distance between your ears will contribute most. I don't know whether this is true, just my conjecture
The phase shifting of the speakers would generate no new problems, as far as equal for all speakers. The same time delay occur by same frequencies.
Ok. Assuming, of course, that the speakers are just drivers and not complete enclosed speakers. The frequency/group delay/phase shift relation in bass reflex enclosures would complicate the problem somewhat, I imagine.
What distinguish the radar from audio is, we can use psychoacoustic rules in addition. As far as we can avoid misguiding cues before the direct wave hits the listener, the success is established by far.
For sure, taking advantage of some of the shortcomings of human perception is extremely useful.
Interestingly, there is also plenty of literature looking at 3D wave field synthesis with ultrasound for medical applications.
Thanks again for publishing this here, I found it an extremely interesting read.
Even handling 300Hz to 3.5kHz is nearly four octaves, so the challenge is large. How much research has been done looking at the frequency bands which are key for direction perception
Yes. We must ensure sufficient extension of the speaker field for the low end limit. On the other hand is not possible to given up the tight spacing of the single speakers because of spatial aliasing. For a highly sophisticated approach some thousands of speakers, each guided by its own amplifier, de- multiplexer and taking a piece of the DSP performance cannot be given up.
But the easy tasks all solved all today. Possibly the effort is an advantage. Why should a producer invest in a technology, if you can buy the products for some bucks on end?
Obviously the low frequencies aren't and the extreme high frequencies can't be (the phase shift is too much), so I would guess that frequencies where the wave length is on a similar order to the distance between your ears will contribute most. I don't know whether this is true, just my conjecture
This assumption is correct. Low frequencies hasn’t considerable phase shifting between the ears, above 3,5 kHz becomes the interaural time differences ambivalent because of the wavelength. It would be an advantage to generate exact conditions also above because amplitude based evaluation occur in that range. But we can tolerate that.
It is amazing yet in what extend we can tolerate such faults. By various ocassions have listen Wave Field Synthesis speaker rows around the listener, build up by 20cm and more spaced speakers. In theory should such alignments cannot work properly above 1 kHz, but in praxis generate such rows already the best spatial impression, I ever able to listen for a reproduction.
The frequency/group delay/phase shift relation in bass reflex enclosures would complicate the problem somewhat, I imagine.
You need DSP assimilation in any case. Low Frequencies undirected air pressure alone. Because you can referring onto a default listener position by my approach, you can fit the time.
Interestingly, there is also plenty of literature looking at 3D wave field synthesis with ultrasound for medical applications.
Lot of applications highly interesting. Sound becomes like shapable material by a wfs- speaker field. I don’t know would be an microphone application for determine the spatial position of a mosquito by its sound by wave field analysis, switch into ultrasound generators by focus point upon the mosquito position for shot it up would be a good application.
Regards Helmut
That was a very interesting read. The English version has some translation problems at times and can be a bit hard to follow, but is generally fairly clear.
Do you have plans for formal publication of your results in a peer reviewed journal?
Hello cabbagerat,
whats about the hydrogenaudio - wiki. Wave field synthesis is empty. By my opinion would be the http://www.syntheticwave.de/Wavefieldsynthesis.htm (http://www.syntheticwave.de/Wavefieldsynthesis.htm) site a good base. Some mistakes in translation I have correct today, I hope. The wiki text should describe more the opinion of the scientists as my own, but the side would be a good base.
Would anyone by native tongue build it together with my ?
Regards Helmut
Even handling 300Hz to 3.5kHz is nearly four octaves, so the challenge is large. How much research has been done looking at the frequency bands which are key for direction perception?
Um, unless I'm very confused, the DL's for ITD are rather larger in that range than they are at both lower frequencies (for waveform) and higher frequencies (for waveform envelope), are they not?
Um, unless I'm very confused, the DL's for ITD are rather larger in that range than they are at both lower frequencies (for waveform) and higher frequencies (for waveform envelope), are they not?
Hello Woodinville
For certain angle regarding the source occur the same ITD for all frequencies of course, as far as you not changing the distance of your ears.
What confused you is possibly my posting in #5, to fit the low frequencies according the upper. That only required for the special case of wave fronts, which align the wave fronts for reflection upon rendition room walls. The procedure utilize such aimed straighten for cheating the listener for a virtually changed listening room size. For example you can delay the ceiling reflection regarding the direct wave for endanger a virtually heightened room perception. See the depiction:
http://www.syntheticwave.de/pictures/principe.swf (http://www.syntheticwave.de/pictures/principe.swf)
On the other hand you can downsize the rendition room spacing of course, if you radiate the early reflections, before the direct wave is generate. That would be interesting for too large build event halls.
Only for such concern would be need a different handling for low and high frequencies for equalize the faults of an insufficiently large speaker field.
H.
Um, unless I'm very confused, the DL's for ITD are rather larger in that range than they are at both lower frequencies (for waveform) and higher frequencies (for waveform envelope), are they not?
Hello Woodinville
For certain angle regarding the source occur the same ITD for all frequencies of course, as far as you not changing the distance of your ears.
What confused you is possibly my posting in #5, to fit the low frequencies according the upper. That only required for the special case of wave fronts, which align the wave fronts for reflection upon rendition room walls. The procedure utilize such aimed straighten for cheating the listener for a virtually changed listening room size. For example you can delay the ceiling reflection regarding the direct wave for endanger a virtually heightened room perception. See the depiction:
http://www.syntheticwave.de/pictures/principe.swf (http://www.syntheticwave.de/pictures/principe.swf)
On the other hand you can downsize the rendition room spacing of course, if you radiate the early reflections, before the direct wave is generate. That would be interesting for too large build event halls.
Only for such concern would be need a different handling for low and high frequencies for equalize the faults of an insufficiently large speaker field.
H.
I am referring to headphone results that test ITD DL's without acoustical involvement.
By the way, it is also incorrect to say that ITD's don't change with frequency for a given angle. Diffraction and boundary-layer issues do actually create some differences there. It is a good question how much those matter, but I do suspect there is some reason to suspectthat these differences help in disambiguation of position, monaurally.
I am referring to headphone results that test ITD DL's without acoustical involvement.
By the way, it is also incorrect to say that ITD's don't change with frequency for a given angle. Diffraction and boundary-layer issues do actually create some differences there. It is a good question how much those matter, but I do suspect there is some reason to suspectthat these differences help in disambiguation of position, monaurally.
Hello Woodinville,
possibly you know newer results, can you give a link? I dont know the boundary- layer- effects.
All what I know in that matter is the huygens- fresnel considering of diffraction effects. According that causing the longer detours of some elementary waves around the listener head only a loss in upper tone range.Those psychoacoustic effects alredy described by Blauert.
But we want to restore the soundfield physically. Forget all Psychoacoustics in that matter.
Regards Helmut
http://www.syntheticwave.de (http://www.syntheticwave.de)
- The procedures you describe are fundamentally very similar to active beamforming as applied to antenna arrays for radar and communications applications. Several excellent beamforming algorithms exist for 2D antenna arrays, and I can't see how they wouldn't work for sound. Have you included antenna beamshaping in your literature search?
Hello Cabbagerat,
I have seen your http://www.brooker.co.za/fers/ (http://www.brooker.co.za/fers/) website, the FERS Project and some other radar and sonar related sites. I am suprised, in what extend the properties similar, thanks for the advice.
In the past days I have improved the english translation of the http://www.syntheticwave.de (http://www.syntheticwave.de) site. Its now a easy understandable description of the wave field synthesis principle, I hope.
Regards Helmut
Hello,
the principle of the wave field synthesis until today is implemented by loudspeaker rows in ring around the listener. The rendition remains two- dimensional therefore. I gladly would discuss my idea for the physical reconstruction of the sound field in all three space dimensions by a speaker wall in front of the listener. The procedure is described by http://www.syntheticwave.de (http://www.syntheticwave.de) . The first site shows an animated and easily understandable description of the WFS principle.
Regards Helmut
New animation for the transformation principle:
http://www.syntheticwave.de/sound%20field%...nsformation.htm (http://www.syntheticwave.de/sound%20field%20transformation.htm)
It shows the transformation from 16*8*6m recording room soundfield into a 6*4*2,7m living room. What you see as remaining problems?
H.
Hello,
the principle of the wave field synthesis until today is implemented by loudspeaker rows in ring around the listener. The rendition remains two- dimensional therefore. I gladly would discuss my idea for the physical reconstruction of the sound field in all three space dimensions by a speaker wall in front of the listener. The procedure is described by http://www.syntheticwave.de (http://www.syntheticwave.de) . The first site shows an animated and easily understandable description of the WFS principle.
Regards Helmut
I want to add a good link for a literature reviev regarding wave field synthesis principle. The site is build of the german Institute für Rundfunktechnik, working in the matter of wfs:
http://www.hauptmikrofon.de/wfs.htm (http://www.hauptmikrofon.de/wfs.htm)
Regards Helmut
Hi, in the last weeks I have depict my proposal for a combination of the impulse response based and model based wfs-approach. Its now on the website:
http://www.syntheticwave.de/wave-field-synthesis.htm (http://www.syntheticwave.de/wave-field-synthesis.htm)
Should that deliver a practicable procedure?
Helmut
Question regarding early reflections
By the model based approach for wave field synthesis become possible to providing the correct synthesis of early reflections for especially dry recorded audio.
Unfortunately hardly records available in appropriated standard until today. My question in that matter seems difficulty: I am of the opinion we can hardly distinguish direct sound from early reflections, as far as both source directions doesn't differ.
If so, which subjective impression should become a normal recorded mono signal, per example the center surround channal, if its audio used for additionally early reflections from different directions?
I cannot find scientific articles in those matter, is only my feeling it would work possibly satisfying. If so, then would become possible in certain degree using conventionally recorded audio for wfs reproduction, placed in suitable virtually environments.
Anybody know sources to spy out the question?
Kind Regards Helmut
http://www.syntheticwave.de (http://www.syntheticwave.de)