I did a little bit more accurate measurement of the "matchbox shaker"; instead of "scientific filter" I used cooledit's "FFT filter" (Hanning, 24000pts). It appears, that there is a 1.5kHz window, between 8000Hz and 9500Hz responsible for vertical axis.
I created two files. Band-pass filter (between 8000-9500Hz only) and band-reject filter (all, except 8000-9500Hz). It looks, that when the small 1.5kHz band is removed, then the sound is comparable to typical binaural recordings (vertical movement does not goes lower than neck/shoulder line). Resonance and/or filtering parts in our head, responsible for vertical hearing - will be then related to this general frequency range. Files combined together via multitrack mixing - give back the whole effect.
Here are they (flac):
Now - I agree, that your measurements can be slightly different than these above. It's about the principle. I was curious, whether the frequency band required - is "all above threshold" or "specific band" related. I also was curious, whether the perception threshold is sharp or soft (in "sotf" case would be difficult to find it).
I'm not claming, that the information about localization is only in this 1.5kHz band, because it can be a matter of interrelations between the lower and higher band. However this window is pointing to something, that probably is vital in practical experimentation.
Another practical conclusion. Next step would be to manipulate this "spatial" band. How much amplitude is required to have the verticality anyway? How much reverb and other distortions can be added without loss of verticality of sound? Such experiments will probably tell - how accurate must be the final design, how immune it is to diffusing distortions.
As an example. The amplitude modification can be -12dB or +12dB, and verticality is still there. Delay less than 3ms does not makes too much difference in verticality. I have not yet found a reverb that will diffuse/distort this concrete 1.5kHz band, but some cheap flanger did not made a change. Such kind of experiments are thought provoking (-;
I am following this discussion for a while. I really wonder whether there is holofinics. The matchbox is nice but can't hear the box in front of me. Behind me is good. This is my holophonic. Here are two takes, just to amuse you during the discussion;
take 1: original stereo
take 2: 3d I'm in the middle of the playgrond
Holophonics is just a word that describes something. We could create another name. Holo-, Spatial-, 3D-, Ambi- /+/ -sonics, -acoustics, -phonics and so on. It's about replicating certain properties of sonic perception. If you hear spatially - then yes, there is spatial hearing. And we use word "holophonic" just because of habit and content association.
Hearing "in front of" is probably more associated with visual context and expectation. When I close my eyes, and forget about my intentional directionality / expectation, I guess I rather hear a lot of sounds behind me, not in front of me. Or so to speak - it is very easy (too easy) for me, to convince my self, that the sound I hear - comes from behind and not the front. Psychology of perception displaces the mechanics of it.
An example from psychoacoustics. When you build a soundspace, and want to push out certain sonic content outside the headphones, you can do this by creating another sonic content in the center of your focus/attention (sensory conflict due specific expectation). It's not so easy in practice, but when you have certain understanding about sound - you can achieve it, and you get the result - only with (proper) mixing, without binaural recording techniques or spatialization/hrtf/reverb plugins at all.
I guess, that proper "holophonic" sound can be achieved with 4 to 6-point assymetry of shadowing environment (filtering and/or resonance factor) which could be be achieved with 1-3 "items" around capsule, plus classic binaural spatialization. Plus perhaps some amplitude related modeling (bigger level differences).
I guess, hearing spatially can be related with proper learning. An example from psychology of hearing. You can not learn a foreign language if you can't hear its content; learning a language has to do with the ability to repeat words, vowels, constructions. Research shows, that different languages operate on different frequency spectras, and that why it is difficult to learn certain languages. If you speak English for example - try to learn German or Polish, or in opposite direction. Do you recognize the structure of what you hear? Our hearing is strictly related to speech, because we live in speaking environment, and must find ourselves there. I can imagine, that mastering methods (for sound) used in US differ from ways of sound production from Germany or Poland, because hearing is different (and they differ according to my professional friends a lot, and it is not a matter of pro or amateur work; different approaches, but in common trends, they say). I recall, that there is a book written by Joshual Leeds, he describes a way for musicians to learn to go beyond limited perception into the depth of sound (they learn to hear, before they can learn to listen); it uses a Tomatis method. I went through my "training" during experimentation, where there was even a time, when I stopped to recognize any music for a few days (everything sounded false and out of tune). (-: So - spatial hearing is related to the learning of hearing, I think.
Definition of OBE changed in last years, or at least - we have several more theories today. One refers to a situation, when a doctor - using a digital robot interface - is operating a patient's heart via internet. Is the doctor out of the body? In certain way - yes. But 40 years ago - this theory would not work, because there was no "advanced" technology like now. People believe that they are "out of body" because of sensory sensations they have, while mechanisms resposnible for that phenomenon - does not have to work that way. Now what is important - phenomenon as a "inner experience" does exist, but around it - we have neuroscience people, newage people, church people, and other kind of movements like that. In such environment - it is difficult to make reasonable and useful theory in terms of practical results. "If I go out of my body, and something eat my OBE spirit - will I die?" Well - it depends on theory you believe in (-; But the question itself, and lack of reasonable science behind - makes many people fear to expand their own potential, and they prefer to agree with their brain in "where the self ends, and the outer world begins".
I wrote about "mimicry" not to ofend you, sorry if sounded to hard. A few days before, I just read some funny articles, in which very similar sentences were used to "limit" a point of view in order to make it suitable to "mainstream theory" (I don't remember details). I'm not against the existing body of research. But if you look around, on many other conversation boards - you will find a lot of people "just arguing", with no practical conclusions; they just exchange theories, taken from somewhere else, and produce "mental/emotional trafic", so to speak. While discussion itself is good, amount of certain trends - decides about it's quality. If a hundert people made dummy head recordings, and achieved certain but unsatisfactory results - I see no reason to stuck where they did; I prefer to try something new, while trying to briefly understand what was wrong with the old design. A lot of major/breakthrough discoveries came from accidental experiments or even dreams, in situations where people were not tied to specific expectations or theories. This system works too pretty well (-;
Out of interest - have you tried a blind test on just swapping L-R channels (and/or reversing headphones) on listeners and then asking them whereabouts the sound is heard?
I did such experiments, but without going into "blind testing" in this area. There is a "research body" on that, made by other people. I focus only on two directions. Deep "vertical axis" and spatial sensations of sound "beyond headphones" so to speak. Somewhat this reduces my search to certain sonic content (able to produce such artifacts), but I need it in my consciousness work. While "far away sound" can be localized even if you feel/hear it "between headphones" (again - sensorial conflict), situations "outside headphones" (when you take off your headphones, to check if the sound does not comes from outside) produce better experience even if localization will be more uncertain/difficult.
Okay, I did an experiment with just one mic. I will post the link. I needed to sign up just to share this.
Here is the link to my example, then I will explain how I did it. I did it with one mic, btw.
Here is how I did it.
First, I made a paper funnel to go around my Microsoft lifechat LX-3000 microphone. Well, it is a USB headset, really. Anyway, I cut the end of the funnel to resemble the shape of the human ear. I sort of smashed the funnel so it was ovalish. The mic sat near the center inside the funnel. The funnel extended about 3 inches away from the mic. One side of the funnel was cut shorter than the other, to resemble the human ear. I can show how I cut it in a youtube video.
Next I recorded without the funnel. I recorded me shaking pills in a bottle at the same loudness and distance from the mic.
Next I set up the mic with the funnel facing left. I sat the headset on a cylindrical object. The funnel faced left, as if it were a riffle barrel facing left.
Then I played the recording through two speakers. One face down so it wouldn't be picked up much by the mic. The other I held in my head and moved in around toward the left of the mic. I moved it up and down, etc.
Then I repeated the process, but turned the mic to point right. I had to flip the funnel to be the right ear as well.
The hardest part was to move the speaker identically to how I did in the first recording, which I couldn't have done perfectly, but it worked.
I want to do this same thing but with two microphones and a single jack, but I have no idea if it will work the same. I have no idea if I could separate left and right from two mics recording from a single jack, using some sort of converter at radio shack. I might just do the binaural headset DYI trick. But I want to do it with the paper cones rather than inserting the mics in my ears, just to see if it sounds better than typical binaural recordings.
Does anyone know how I can get two separate channels L R with two mics and a single jack?
Experiment no. 1. Rectangular room with poor acoustics, OKM binaural mics , poor matchbox and me. And my first test of these binaural mics at all, because just received them. Listen with headphones, although at the end of the recording - vertical axis is perceivable even in 1 earphone. Perception confirmed by more than 2 independent listeners (with different ear geometry).
It looks, that the vertical axis is easy to achieve.
UPDATE. inside the zip - there is the reference holophonic file for the pool question:
Okay. In the next days I will do some more experiments, to check certain concepts - how to increase the effect.
The last 10 seconds is vertical shake (from up to very down). Everything earlier - is moving somewhat only around (I had the recorder in one hand, and poor matchox in another - had nothing else to use, and I was wierd by cables, so it was a little difficult to navigate).
It's odd and interesting to me, that you did not heard/recognized the vertical movement. Several people from my side who tried this, on various headphones (ranging from "cheap" 10 USD to "expensive" like 200 USD) - they perceive it very clearly. Some of them have even hearing difficulties (high frequency band limitations due to age related degradation of hearing). On the other hand - I don't hear it in front of me (but I don't remember whether I shaked it in front of me; I focused probably on "a little behind" verticals) (-;
@Microscopia - let me ask you some questions. Did you used classical headphones setting (2.0)? Did you hear the vertical movement in the reference "matchbox shaker" file - using the same listening equipment? Can you exclude interference from DSP effects (software or hardware based, including EQs) or effects related to mechanical construction of the equipment, headphones for example? (in other words - different playback output, different headphones, something more ordinary). I assume that with your hearing and/or general sound localization is fine (you don't need to answer that one)? The last question comes from the fact, that some people can have certain difficulties with proper sound localization not only because of physical hearing damage, but also due neurological issues (like facial recognition problems which I have - I am able to not recognize familar faces as faces of people I know) or because improper auditory learning in the past (so, the person can hear it, but does not knows what it means).
According to some previous measurements (see a few posts before), affecting certain frequency bands (at least one) with amplitude change or certain degree of sound diffusion - will limit the vertical perception to neck/shoulders level. I haven't measured yet my recording to check if the parameters are similar, but I will do so.
I added a small pool. If some questions should be redefined or added, let me know.
I just realized, that the dog I hear barking - is not behind the window, but in the recording But I came to something, that can be useful durig listening tests. Loudness. At certain (higher) loudness - I would describe this recording as "generally average quality". Better headphones usually provide better dynamics with greater loudness. It reminds me also, that we have non-linear loudness perception, so the "decibel curve" in frequency domain changes, depending on how loud (and far from natural) we listen. While with clean, studio recordings, there is no such problem - in home made shots, with background ambiance - there is a big difference at different levels. I don't know if this can affect the spatial perception too much, but try to listen so loud as if it was a real matchbox nearby your head. It will limit the noise floor and sharpen sonic details.
Here are a few thoughts after some short experiments with OKMs and human head (unfortunately I can't upload these examples, because in order to navigate the samples later - they are commented aloud). I have not began yet any experiments with additional sound-shaping accessories (-;
1. Vertical localization seem to depend on distance, how close to the body is the sound source. Because I had nobody to help - I don't know yet where is the "boundary" of good/bad recognition.
2. Vertical localization seem to be the best on sides of the body instead of front/rear positions. Somewhat the boundary line corresponds to the line of external ears.
3. Vertical localization seem to depend on timeline context. Sound is better localized down for example, when it moved from somewhere else above. When it is only down, then the perception is confused and position "switches" between places. So it depends on expectation, but this can be shaped by movement.
4. With current setting - I don't hear the recorded sound in front of me (only behind me), but... certain tonal instruments (like gongs) I hear in front of me only.
Next thoughts - later.
some fun: youtube
@Microscopia - just an idea came to my mind. If you don't hear the vertical movement in last 10 second shaking periond, but you hear horizontal spatial movements, perhaps you can try to describe something. Do you hear any difference in sound - between preceding section and the last 10 seconds section? I don't know, flanger, equalizer that moves up and down (brighting or darkening the sound or making it louder/quieter), things like that perhaps. It would help to distinguish the nature of the phenomenon.
Another freesound user just said (see comments section), that can't hear verticality (still it is about 10% of subjects; I had another positive confirmations on my side, among my contacts - only one is negative). But what is interesting - user says, that does not perceives vertical movements on both files - mine and of Zuccarelli. If it was not a "friendly provocation" (my trust to this fellow is somewhat limited because of his past actions...), this tells perhaps something about the origins of Zucarelli's approach (standard method + marketing) - no matter whether there are perceptual and/or hardware issues involved.
I do some checkups tomorrow. I hope that I will find someone who will help me to record more samples from different perspectives (I'm not able to do it alone), that will be great reference (I'd like to check some mics positions and several more ideas).
There is one thing you mentioned. In fact, when I "switch" my vertical perception off (distract enough to lose the track where sound source goes) and then recall it back, there is a moment, when I hear the sound much more far away from me than it should be (it's literally a fraction of second). Then the low placed vertical position goes back, and the sound source is close to me again. I also found an interesting thing. No matter whether I sit, stand or lie down, when I'm in straight position (head-legs portion of my body) - then everything is fine. But when I sit cringed (making about 90 degree bow line), then vertical placement of sound tends from time to time to escape into "horizontal and far away from me" as you describe (it's probably due to sensorial conflict between body perception and sonic expectation).
By the way. In reference zip file, there are "holophonic", and "stereo" files of the Zuccarelli's matchbox. Did you compared their spectras? Did you tried to compare holophonic reference with my matchbox? If holophonic and stereo references are similar, then such spectral measurements will probably fail and things like "reverberations" will get in place. But if not - I'm curious if my matchbox differs much from so called "holophonic".
from the internet via google:
Normally, the apparent position of a sound source corresponds closely to its actual position. However, in some experimental situations listeners make large errors, such as indicating that a source in the frontal hemifield appears to be in the rear hemifield, or vice versa. These frontback confusions are thought to be a result of the inherent ambiguity of the primary interaural difference cues, interaural time difference (ITD) in particular. A given ITD could have been produced by a sound source anywhere on the so-called "cone of confusion." More than 50 years ago Wallach [J. Exp. Psychol. 27, 339368 (1940)] argued that small head movements could provide the information necessary to resolve the ambiguity. The direction of the change in ITD that accompanies a head rotation is an unambiguous indicator of the proper hemifield. The experiments reported here are a modern test of Wallach's hypothesis. Listeners indicated the apparent positions of real and virtual sound sources in conditions in which head movements were either restricted or encouraged. The frontback confusions made in the restricted condition nearly disappeared in the condition in which head movements were encouraged. In a second experiment head movements were restricted, but the sound source was moved, either by the experimenter or by the listener. Only when the listener moved the sound source did frontback confusions disappear. The results clearly support Wallach's hypothesis and suggest further that head movements are not required to produce the dynamic cues needed to resolve frontback ambiguity.
I will need to try with these small head movements, because no matter what I do - on final recordings - I can't hear it in front of me, only rear Although from time to time I have a glimpse of frontal position. Rigid neck/head (-;
another useful quote from the internet (about measured differences between binaural settings):
Three groups of hearing-impaired listeners who had been filled binauraily with behind-the-ear (BTE). in-the-ear (ITE), or in-the-ear-canal (ITC) hearing aids were tested on spatial localization function for sources in the frontal horizontal and vertical planes. No significant differences in unaided performance were observed between the groups, nor between that and aided performance in the BTE and ITE wearers. ITC wearers, by contrast, showed a deterioration in aided over unaided performance. From observations of performance when each group wore temporary fittings of the other two types of system, and from the performance of a non-impaired control group, it appears that the performance decrement for the ITC wearers was due to their own particular systems although specific reasons for this decrement could not be identified. In all conditions, aided and unaided, vertical plane localization was markedly disrupted in all the hearing impaired groups. It was also disrupted, to a lesser but still substantial extent, in aided conditions for the non-impaired listeners.
this link seems to be interesting to visit:
Learning Sound Location from a Single Microphone
my ears don't hear much vertical effect in your sample,a bit from top of head to ear but basically nothing from ear to feet.... very similar to my own results recording with binaural mics
.... The Zuccarelli matchbox has suprising verticality although it is out on the sides....out on the sides because there is very little sound in one channel at the moments of greatest vertical shift.Gradually increasing the volume in the quiet channel neutralises the rising/falling effect as it adds increasingly a backwards/forwards element,which is distracting...so,like you,i have yet to hear a good solid sound falling straight in front of me and assume that such a sound,if possible, will need more strong signals in both channels.
Thanks That's an interesting spectrogram,it covers the best bit of the sample and by looping playback to fit the image i can confirm that the shift in vertical axis follows the V's (up and down) exactly.....so there's a whole lot of eq shift going on here!
It would seem that the perception of vertical axis runs close to the limits of the ears capabilities so i hope someone who can will write an algorithm we can all use to reproduce this effect and improve our hearing.
.........angels descending anyone?
@juskiddink - thanks for comments.
Well - I think - now is time to begin the REAL experimentation on - how to make it from binaural to more realistic. (-;
On one hand - physical design modifications/adjustments, on the other hand - offline sound manipulation. I'm getting an impression, that in Zuccarelli's recording, certain (and audible/recognizable) aspect of sound is just louder than in mine. I could be wrong, but perhaps this is a cue.
Before I forget. I did similar EQ band filtering on my file like on the reference matchbox earlier. There is one interesting point. On the reference file - when I removed this 1.5kHz specific band - deep verticality disappeared totally. But after removing similar band in my recording - only a small part of verticality was gone (c.a. 20-30cm plumbline/segment on the shoulder level), and the sound just popped down.
I just did some measurement on channel separation (left versus right). During up/down shaking period - max peak levels (not RMS) are similar in Zuccarelli's and in mine, c.a. 18-19dB. When sound is down - this separation is c.a. 12dB, but still - similar on both recordings.
I also did side by side comparison of both recordings. Zuc' matchbox shaked on my left, my portion - on my right. Well - I have to change my matchbox to better model (-:
While I perceive similar vertical depth on both, there is one difference (but I don't know yet if this is due to poor matchbox I used). On Zuc' recording I hear deeper EQ as it moves in frequency domain, while my recording is more flat and somewhat difused. On software level it would mean, that if a plugin is able to track and affect such trench, then perhaps it will be able to equalize binaural recording into something more. Perhaps perhaps (-:
I moved with my experimentation here:
From what I understood - HRTF is a rather dead direction in holophonic(s), and it is one of the "hot spots" of many THEORETICAL discussions on the internet. I don't care what is the mathematics behind the "vertical axis" (-;
Friends - ONLY sonic examples of "vertical axis" (and achieved depth) will speak for themselves. We should try to avoid "scientific theories", and focus on facts that we can test and measure. Theory will make feel you better, but only facts will make you hear better. We could spend months on talking who is right. Very quiet and silent months, with no sound around... (-
Sorry I am a little late to the party.. Try http://www.freesound.org/people/dwareing/sounds/169373/
I would love to know if it works for anybody other tham myself!
As for the bottom part - how close to the "vertical body line" did you recorded? I mean - if a "person" is standing, and you play around the body space - how far from it?