Same protocol, same measuring program, and absorbing dummy head (but too flat for having a good separation)
The time scale is not the same, there's a looot of noise, but I think that basically these plots show the same events, like two reflections peaks, for me at 15 and 21 ms.
But CTC is really absent in my plots. Next time I will try with a volunteer's real head.
The time scale is not the same, there's a looot of noise, but I think that basically these plots show the same events, like two reflections peaks, for me at 15 and 21 ms.
But CTC is really absent in my plots. Next time I will try with a volunteer's real head.
Attachments
Yes it will be very narrow. Cross talk cancellation provided by the box is too little to have correct ILD for amplitude panned stereo. There were real numbers in Blauerts book of how much of ILD is required for a specific amount of lateral angle phantom shift, but I don't have time to check it now..
- Elias
Radugazon, Nice dummy 😀
About the plots.. There is a 'feature' in the Octave wtf package that it normalises the maximum of every wavelet plot to 0dB, so comparing absolute values between two plots is impossible. I removed the normalisation and it plots absolute values now (see my plots dB scale is -50 to -20 dB).
Soon there will be a new major release of the package including uncountable number of new features 😀 I'll inform in the wtf thread.
- Elias
About the plots.. There is a 'feature' in the Octave wtf package that it normalises the maximum of every wavelet plot to 0dB, so comparing absolute values between two plots is impossible. I removed the normalisation and it plots absolute values now (see my plots dB scale is -50 to -20 dB).
Soon there will be a new major release of the package including uncountable number of new features 😀 I'll inform in the wtf thread.
- Elias
are those measurements taken on-axis? that is an important question because back-to-back is a wide sweet spot solution - the sound picture remains basically the same for a listener sitting off-axis - or at least it was the case of my 23x23x23 cm cube test box with two 4-inchers
therefore my interim conclusion is that back-to-back cannot rely on crosstalk reduction in the direct sound
For the same reason it seems to me that it doesn't rely on any specific reflection pattern either.
Elias, Scott - can You confirm this off-axis image stability in case of Your test boxes?
I am not sure whether what is going on can be called summing localisation in the conventional meaning of the concept
by localization You mean relative left-right positioning and depth?
in extreme nearfield only or also at some distance from the speakers?
Have You measured them outside? What is the frequency content of the direct sound?
I wonder how much of it is really required? Taking into account the fact that channel separation required for intensity stereo to work could be as little as just 6 dB:
DPA Microphones :: XY Stereo
but isn't how much of ILD is required for a specific amount of lateral angle phantom shift a question different from how much of channel separation is required for intensity stereo to work?
I wonder if when back-to-back is in a normal, more or less reverberant room, our hearing in the presence of characteristic set of reflections can interpret the situation coherently and does it by simply disregarding the direct sound qua the source of the first wavefront determining the direction of sound?
After all hearing is about processing information and between the first wave arrival at the ears and the initial phase of a sound source identification (human being becoming aware of a particular, identified sound source) 250 ms passes which is also amount of recordable time in echoic memory; that is, chunks of sound stimuli are recorded in echoic memory at this length of time
From the perspective of our conciousness this is very short time, we are simply not aware what is going on until after 250 ms therefore we think that we can identify a sound source immediately but from perspective of acoustics of a small room this is a whole lot of time
moreover the time needed to cortically process musical elements is 100 ms so the brain can take many samples before the cortical process leading to a conscious experience really starts at all
How big was the soundstage? 1' or even wider? Did you also perceive added depth or did the speaker distance define the sound distance?
Hello,
About the localisation of phantoms with cardboard we should consider the experienced fact I did with Markus' sound demo samples when I added artificial stereo reverb, namely with reverb the localisation was very blur however the sound was very nice more realistic than the pure sound samples alone. And remember that those demos were only ITD, but no amplitude panning. My conclusion of this is that with reverberation the ITD becomes ambiguous and cannot be relied upon with generating phantoms with the cardboard.
Also as indicated before with my binaural measurements the ILD of the direct sound is very small, too small to generate phantoms.
Thus the cardboard seems to rely totally on lateral wall reflections in forming the phantom images.
- Elias
About the localisation of phantoms with cardboard we should consider the experienced fact I did with Markus' sound demo samples when I added artificial stereo reverb, namely with reverb the localisation was very blur however the sound was very nice more realistic than the pure sound samples alone. And remember that those demos were only ITD, but no amplitude panning. My conclusion of this is that with reverberation the ITD becomes ambiguous and cannot be relied upon with generating phantoms with the cardboard.
Also as indicated before with my binaural measurements the ILD of the direct sound is very small, too small to generate phantoms.
Thus the cardboard seems to rely totally on lateral wall reflections in forming the phantom images.
- Elias
panostereolithically ?
I have been testing all the day a variant of Elias's idea.
The front speakers are my usual line array of linear quadripoles, side firing "à la stereolith",. Their radiation pattern is omni but with a reinforcement on the drivers axis.
The back speaker is the black winged thing. It's sub is actually in front, then the whole system has 3 subs, two dipoles and one bipole, all equidistant from the listener. The back-front space is 6 meters.
The adjustment of the levels is done by resistors, a security as all the drivers are paralleled. All this makes 22 drivers running together with 2 kVA and 0.5 Farad in the PSUs.
The difference with Elias's is that the back speakers being very directional, it's a temptation to swap the channels and put them in reverse polarity. (that's why the boomers are moved in front, to not create any cancellation).
If the listener is in the center, if the reflection paths are the same...little CTC.
But it's not so simple to have a balanced FR and to found "the" point. Work in progress. The wavelets are not very clear. Measured with my own head of dummy, mike in the ears. Maybe there is a controlateral attenuation under 1000 Hz. Better to wait the new release.
Listening is much easier. The optimal spot is for one head (Np, I have only one). I have removed the lateral diffusers (now in the front corners).
The result makes me very very happy. Not only the imaging, the spaciousness, the absolutely 180° display, the immersion are what you can suppose, but the quality of the timbres, the abundance of details are also improved.
Et voilà...
I have been testing all the day a variant of Elias's idea.
The front speakers are my usual line array of linear quadripoles, side firing "à la stereolith",. Their radiation pattern is omni but with a reinforcement on the drivers axis.
The back speaker is the black winged thing. It's sub is actually in front, then the whole system has 3 subs, two dipoles and one bipole, all equidistant from the listener. The back-front space is 6 meters.
The adjustment of the levels is done by resistors, a security as all the drivers are paralleled. All this makes 22 drivers running together with 2 kVA and 0.5 Farad in the PSUs.
The difference with Elias's is that the back speakers being very directional, it's a temptation to swap the channels and put them in reverse polarity. (that's why the boomers are moved in front, to not create any cancellation).
If the listener is in the center, if the reflection paths are the same...little CTC.
But it's not so simple to have a balanced FR and to found "the" point. Work in progress. The wavelets are not very clear. Measured with my own head of dummy, mike in the ears. Maybe there is a controlateral attenuation under 1000 Hz. Better to wait the new release.
Listening is much easier. The optimal spot is for one head (Np, I have only one). I have removed the lateral diffusers (now in the front corners).
The result makes me very very happy. Not only the imaging, the spaciousness, the absolutely 180° display, the immersion are what you can suppose, but the quality of the timbres, the abundance of details are also improved.
Et voilà...
Attachments
Final attempt to prove that the direct sound of cardboard has no meaning in generating phantoms. Graduation picture ! (maybe first grade.. but there is the hat !)
Added acoustic absorption panels on the both side to kill the side signal. Only direct sound is present.
The center element is mute, only side elements are active.
When I amplitude pan 100% from side to side i.e. all the signal goes to one element only, the image stays between the panels. The image can be shifted +/- 20 cm when listening at 2 m distance. Useless !
It is the lateral reflections this cardboard depends on.
- Elias
Added acoustic absorption panels on the both side to kill the side signal. Only direct sound is present.
The center element is mute, only side elements are active.
When I amplitude pan 100% from side to side i.e. all the signal goes to one element only, the image stays between the panels. The image can be shifted +/- 20 cm when listening at 2 m distance. Useless !
It is the lateral reflections this cardboard depends on.
An externally hosted image should be here but it was not working when we last tested it.
- Elias
You are great Elias! 😀
I think that we can safely assume that it
- works through reflections and
- when the listener is well into the reverberant field of the room ie. at a distance form the box >> reverberation distance
which is not that difficult condition to meet in a real listening room because if we take:
and that
then
more on this with sources here:
http://www.diyaudio.com/forums/multi-way/121893-spl-targets-speaker-design-10.html#post1498675
the question remains does back-to-back require/prefer a special pattern of reflections to work best
in other words - how much is type of sound screen - that is the room/speaker arrangement with it's particular reflections patterns (placement of the speaker in the room included) - critical for such a sound projector to work at all and to work best
22 drivers running together with 2 kVA and 0.5 Farad in the PSUs!!!!
this is serious! 
😀no active signal processing at all?
I wonder how it compares to full ambiophonic setup with processing...
all great but then... the BIG question remains!
is 'flat' correct for such a system?!
I'd like to confirm that finding. Image collapses when putting porous absorbers next to the speaker.
High frequency signals also collapse to the center when using a single front-facing tweeter like in the Stereolith.
yes, it is a sound projector like I've been saying for quite a while
question is - does it require a special screen and positioning WRT that screen?
question is - does it require a special screen and positioning WRT that screen?
I've been pondering with the tweeter..
Front facing single tweeter not working, image is in the box.
Best single tweeter so far is between ceiling firing and 45 angled on the top of the box, crossed minimum around 7 kHz. The problem with ceiling firing tweeter is tonal balance changes when I move to the far end of the room. It is also required to block the direct sound from the tweeter.
I also boosted the treble of side firing FRS8 and it improves things quite a bit. Maybe better than with single tweeter.
Propably best tweeter for the cardboard is two side firing tweeters. Or using high quality fullrange with boosted top end.
Tests continue...
- Elias
Front facing single tweeter not working, image is in the box.
Best single tweeter so far is between ceiling firing and 45 angled on the top of the box, crossed minimum around 7 kHz. The problem with ceiling firing tweeter is tonal balance changes when I move to the far end of the room. It is also required to block the direct sound from the tweeter.
I also boosted the treble of side firing FRS8 and it improves things quite a bit. Maybe better than with single tweeter.
Propably best tweeter for the cardboard is two side firing tweeters. Or using high quality fullrange with boosted top end.
Tests continue...
- Elias
Yes, for a wide sound stage the tweeters need to fire to the sides.
With the FRS I had to boost the treble quite a lot (+11dB) to get decent direct sound but the speaker is now way too bright. The solution is most likely a waveguide.