Elias, I don't know why you'r so focused on HRTFs. Localization due to interaural time and level differences is much more stronger especially when talking about stereophony where phantom sources do only appear in a straight line between the two speakers. If there are phantom sources at other positions then HRTF-like distortions might be the cause but this is just an effect and nothing that can be reproduced reliably with every listener because HRTFs are highly individual like a fingerprint.
Best, Markus
Best, Markus
Hello,
Interesting. I would like to know more about your 'ambiopoles' if possible. Is it like stereolithe? Do you have one or two tweeters, do(es) they (it) point towards the listener or side wall? How about midrange?
IF you would make a speaker like this by putting one tweeter pointing to listener and aiming midrange to side wall to reach the listener by wall bounce, this would be quite good solution actually, at least in theory. Why I say so is because in this way there will be no pinna error to make contradictions, so localisation is based on midrange which is reaching the ear at high angle from the wall reflection so at midrange the system works assentially as stereo and provides phantoms. But this is IF, since I don't know how did you realise your speakers.
- Elias
graaf said:
Interesting. I would like to know more about your 'ambiopoles' if possible. Is it like stereolithe? Do you have one or two tweeters, do(es) they (it) point towards the listener or side wall? How about midrange?
IF you would make a speaker like this by putting one tweeter pointing to listener and aiming midrange to side wall to reach the listener by wall bounce, this would be quite good solution actually, at least in theory. Why I say so is because in this way there will be no pinna error to make contradictions, so localisation is based on midrange which is reaching the ear at high angle from the wall reflection so at midrange the system works assentially as stereo and provides phantoms. But this is IF, since I don't know how did you realise your speakers.
- Elias
Hello,
Yes, you are most correct. I also see often references with smallest notable time difference between the ears being in order of 20us.
Also I don't know why these things are not more widely known. Maybe people read too many hifi magazines instead of scientific publications
- Elias
WithTarragon said:
Yes, you are most correct. I also see often references with smallest notable time difference between the ears being in order of 20us.
Also I don't know why these things are not more widely known. Maybe people read too many hifi magazines instead of scientific publications
- Elias
Hello,
Not so. Do some more reading. From Blauert Spatial Hearing there is case with two loudspeakers in s stereo triangle. Freq range is from 100Hz-10kHz. See, there is practically NO change in localisation accuracy when freq is getting lower. This is the results for phantom image localisation blur. The blur is very small, only about couple of degrees! Who tells you human cannot localise sounds at low freqs?!?
- Elias
gedlee said:
Not so. Do some more reading. From Blauert Spatial Hearing there is case with two loudspeakers in s stereo triangle. Freq range is from 100Hz-10kHz. See, there is practically NO change in localisation accuracy when freq is getting lower. This is the results for phantom image localisation blur. The blur is very small, only about couple of degrees! Who tells you human cannot localise sounds at low freqs?!?
- Elias
Attachments
Elias said:
I didn't use separate tweeters pointing to the listener
I use just fullrange units pointing sideways - left and right - first KEF Uni-Q coincidental units and then medium sized fullrangers
so it seems that (of course if I understand pinna error problem correctly?) in fact there was no significant pinna error taking into account directivity of Uni-Q and 10-cm fullrangers - very little high frequency sound reached the ears directly
the sound was as if "projected" on side-walls and reflective wall behind the speakers - I positioned them close to wall, even on the wall with good results
but I tried also plcement in the corner or diagonally away from the walls and symmetrically away from the walls
all placements work well with exception of close corner placement
I have tried different box sizes and proportions
Best results I achieved with small cubic box appr. 25/25/25 cm IIRC but I also use rectangular box of appr. 15/15/75 (the last is length - distance between the speakers) with good results
I remember that 30/30/10 box (speakers as close to each other as possible - magnet-to-magnet) was not good
best!
graaf
Everyone should read the paper that was posted. It is very clear about what the "principle" mechanisms for localization are and what is being stated here is not in line with waht was found in that paper. I posted the pertinent parts but people seem to be ignoring it.
Pinna localization is a HF effect NOT a LF one and loudspeaker directivity at LF will have no effect on this.
Pinna localization is a HF effect NOT a LF one and loudspeaker directivity at LF will have no effect on this.
Hello,
Very interesting topic.
The rest I disagree. Sound localisation in small rooms do not 'must vanish' but instead we NEED TO DO something about it to improve the situation. The solution is to use high directivity sources in the freq range below, say, 1kHz. This is essential!
- Elias
gedlee said:
Very interesting topic.
The rest I disagree. Sound localisation in small rooms do not 'must vanish' but instead we NEED TO DO something about it to improve the situation. The solution is to use high directivity sources in the freq range below, say, 1kHz. This is essential!
- Elias
Hello,
Now we are talking! We are almost reaching the point. But wait, not to jump to conclusions so fast! High freqs will rule in spatial hearing with *real sound sources*, but NOT in phantom imaging! High freqs (>3kHz) are the _worst_ to form a phantom.
Solution is indeed to get high freqs 'right', but the solution is not obvious, and certainly not the one commonly seen in loudspeakers. I'll come to this later.
- Elias
gedlee said:
Now we are talking! We are almost reaching the point. But wait, not to jump to conclusions so fast! High freqs will rule in spatial hearing with *real sound sources*, but NOT in phantom imaging! High freqs (>3kHz) are the _worst_ to form a phantom.
Solution is indeed to get high freqs 'right', but the solution is not obvious, and certainly not the one commonly seen in loudspeakers. I'll come to this later.
- Elias
gedlee said:
Well, I guess you can lead a horse to water but you can't make him drink.
The field is much larger than the paper you are referring to. I have offered up some other things and you are ignoring them.
Once again I will give up on trying to bring you up to speed. It is a very large topic with the associated terminology, techniques, etc, and large set of findings. In all fairness you are currently in a position of simply speculating. If you don't want to listen to me, fine. But I would suggest that you at least talk to someone who publishes on this topic in a journal such as JASA. You can hear it from them.
-Tom
Hello,
Apparently you haven't read the paper, have you.
The quote from the paper is correct in it's content. Your conclusions is all wrong.
I'll quote the paper:
"In the present study the competing sounds were broadband noises, and subjects were asked to simply indicate whether they perceived sound arriving from a single location or from two distinct locations"
One must realise this paper deals with differentiating spatially a *real sound source* in the presence of another disturbing *real sound source*. In that light "separation of concurrent sounds at the frontal location was not impaired by high-pass filtering the signals" is true because high frequencies will and they do locate the source because of pinna! If you want to bring this result to loudspeakers, it means pinna localisation will locate the speakers itself, NOT phantom images.
Quote: "that high-frequency cues can also be sufficient for identifying that more than one location is present in a multiple-source stimulus". This is what will happen if source contains high freqs. It makes the source itself localisable! This is NOT what we want, we do not want to localise the speakers.
Please understand this.
- Elias
gedlee said:
Apparently you haven't read the paper, have you.
The quote from the paper is correct in it's content. Your conclusions is all wrong.
I'll quote the paper:
"In the present study the competing sounds were broadband noises, and subjects were asked to simply indicate whether they perceived sound arriving from a single location or from two distinct locations"
One must realise this paper deals with differentiating spatially a *real sound source* in the presence of another disturbing *real sound source*. In that light "separation of concurrent sounds at the frontal location was not impaired by high-pass filtering the signals" is true because high frequencies will and they do locate the source because of pinna! If you want to bring this result to loudspeakers, it means pinna localisation will locate the speakers itself, NOT phantom images.
Quote: "that high-frequency cues can also be sufficient for identifying that more than one location is present in a multiple-source stimulus". This is what will happen if source contains high freqs. It makes the source itself localisable! This is NOT what we want, we do not want to localise the speakers.
Please understand this.
- Elias
Tom, bad exit strategy. Even Toole didn't find any utilizable data in his recent review of the available scientific papers. And I think he has read a lot of papers in his life. We're talking about small room acoustics. The only thing that is proven (again look at Tooles paper for references) is that all the work that has been done on large room acoustics can't be transferred to a small room!
And have YOU ever searched for "Geddes" in the JAES library
Best, Markus
P.S. What specific paper of Ben Bauer are you referring to?
And have YOU ever searched for "Geddes" in the JAES library
Best, Markus
P.S. What specific paper of Ben Bauer are you referring to?
Hello,
One more interesting point from the paper linked before.
Quote:
"3.8.5 Ineffectiveness of spectral cues in a mixed signal
It was interesting, although not altogether surprising, to find that pairs of broadband
noises could not be resolved when separated along the median vertical plane (with no
binaural differences present). This indicates that the location-dependent spectral cues
of the individual sources were not accessible in the mixed signal under these
conditions."
Meaning if you place high freq sources in or very close to the frontal median plane, high freqs emitting from these sources do not disturb other localisation cues that may take place at the same time.
One famous implementation of this: Ambiophonics.
- Elias
One more interesting point from the paper linked before.
Quote:
"3.8.5 Ineffectiveness of spectral cues in a mixed signal
It was interesting, although not altogether surprising, to find that pairs of broadband
noises could not be resolved when separated along the median vertical plane (with no
binaural differences present). This indicates that the location-dependent spectral cues
of the individual sources were not accessible in the mixed signal under these
conditions."
Meaning if you place high freq sources in or very close to the frontal median plane, high freqs emitting from these sources do not disturb other localisation cues that may take place at the same time.
One famous implementation of this: Ambiophonics.
- Elias
WithTarragon said:
Tom - I really don't think that I disagree with you, not as much as I do with Elias (I don't get what pinna localization has to do with LF directivity.) But I do fail to see, and in complete agree with Toole, how one applies the work of Blauert, etc. to the question of LF directivity in a small room. Beyond that question I claim no expertise. But all the work that I have seen says that the localization ques are dominated by the high frequencies NOT the low frequencies. This does not say that we cannot localize at LF, only that the HF cues of a complex signal will dominate. In this situation it is imperative to get the HF directivity right (to get a clean reflection free field) far more so than the LF directivity. I have seen nothing to convince me that this is not true.
The fact that it is impractical to get LF directivity is another issue altogether.
P.S. I will have to look up the Ben Bauer papers. I was at Penn State while he taught there and I had a great deal of respect for him. I also did some work with him at CBS labs. He was a rare researcher in the field of audio.
The work from Bauer that I am referring was published in JASA in the the early 1960s and would have "stereophony" in the title. The mention of the any stuff about quadruphonic is unknown to me and would probably be off-topic here. I would not dismiss his work off-handedly.
The basic finding (and do not start criticizing until you have actually read and digested the material) is that a "phantom" source is in fact "identical" to a what would be created via stereophony (using two speakers). He provided a phasor analysis (if I recall correctly) for the physics behind this.
The other topic you guys mentioned was localization and localization with echoes/reverberation etc. The rules don't fundamentally change regarding the role of interaual cues.
However there are other issues about the relative contribution of the interaural cues in the secondarily arriving sound (call it an echo if you want). How this "other stuff" is used / combined/ weighted / etc is a topic of considerable interest and research. You may find it is difficult to read up on since it can go by the various names (of perhaps similar phenomena) such as precedence effect, law of the first wavefront, Hass effect, Clifton effect, onset dominance, Fransen effect, and the list goes on. The measurement techniques will differ, so you need to be careful before lumping them all together (a mistake that I just made ....)
-Tom
The basic finding (and do not start criticizing until you have actually read and digested the material) is that a "phantom" source is in fact "identical" to a what would be created via stereophony (using two speakers). He provided a phasor analysis (if I recall correctly) for the physics behind this.
The other topic you guys mentioned was localization and localization with echoes/reverberation etc. The rules don't fundamentally change regarding the role of interaual cues.
However there are other issues about the relative contribution of the interaural cues in the secondarily arriving sound (call it an echo if you want). How this "other stuff" is used / combined/ weighted / etc is a topic of considerable interest and research. You may find it is difficult to read up on since it can go by the various names (of perhaps similar phenomena) such as precedence effect, law of the first wavefront, Hass effect, Clifton effect, onset dominance, Fransen effect, and the list goes on. The measurement techniques will differ, so you need to be careful before lumping them all together (a mistake that I just made ....)
-Tom
Originally posted by Elias I would be happy even for ONE stable phantom source under reverberant conditions
I still think that something's wrong with your setup if you've never got to hear a stable phantom sound source. Are you sure that one speaker or a single chassis isn't connected inversely?
Nearly every recording I listen to shows very stable locations of instruments.