From Griesinger's homepage
12/15/19 In other news…
Matthew Neil and Michelle Vigeant presented a paper at the ICSA conference in Amsterdam describing the results of their tests for the preference of the sounds in a large number of concert halls. The experiments were carefully done, using third order Ambisonics to reproduce the sound from a large electronic orchestra. Separate impulse responses from each loudspeaker were recorded at various listening positions using a 32 channel Eigen Mike. The recorded data was reproduced in an anechoic chamber using independently recorded anechoic music for each instrument.
The result of the preference tests was surprising. At the presentation Neil claimed that seventy-five percent or more of the preference could be explained with a single perception – Proximity. The conclusion is not unexpected from my work – but was I was startled by the magnitude of the difference between proximity and all the standard descriptors of hall sound. Neil exclaimed “but there is no measure for proximity!” I pointed out that there is such a measure, LOC, that uses a binaural impulse response as an input. The measure has been published in Leo Beranek’s last three JASA papers and he gives our original formula for calculating it. (The newer code is better…)
The result was startling to me, not for the fact that people like proximity, but for the degree to which they seemed to prefer it. It is hard to get 75% or more of people to agree with anything! But in the preprint for the talk Neil et al describe that their test subjects were all classical musicians, and not simply picked off the street. The test material was classical, and the subjects knew how to listen to it. I think the selection of the subjects was appropriate. Classical music when well played and well heard is complex, emotive, and fascinating.
I have devoted much of my life to learning how to make and play back classical recordings that hold attention in this way, and learning how to design halls which can provide an even better experience for the majority of the audience. The papers on this site are intended to help other people understand some of the things I have learned. They are primarily preprints or power points, and are not published in standard journals. Hopefully this will change.
http://www.davidgriesinger.com/The_Physics_of_auditory_proximity.pptx
By the way Griesinger designed the only multichannel decoder that Toole likes - Lexicon MC-12 https://www.harman.com/documents/HowManyChannels_0.pdf
12/15/19 In other news…
Matthew Neil and Michelle Vigeant presented a paper at the ICSA conference in Amsterdam describing the results of their tests for the preference of the sounds in a large number of concert halls. The experiments were carefully done, using third order Ambisonics to reproduce the sound from a large electronic orchestra. Separate impulse responses from each loudspeaker were recorded at various listening positions using a 32 channel Eigen Mike. The recorded data was reproduced in an anechoic chamber using independently recorded anechoic music for each instrument.
The result of the preference tests was surprising. At the presentation Neil claimed that seventy-five percent or more of the preference could be explained with a single perception – Proximity. The conclusion is not unexpected from my work – but was I was startled by the magnitude of the difference between proximity and all the standard descriptors of hall sound. Neil exclaimed “but there is no measure for proximity!” I pointed out that there is such a measure, LOC, that uses a binaural impulse response as an input. The measure has been published in Leo Beranek’s last three JASA papers and he gives our original formula for calculating it. (The newer code is better…)
The result was startling to me, not for the fact that people like proximity, but for the degree to which they seemed to prefer it. It is hard to get 75% or more of people to agree with anything! But in the preprint for the talk Neil et al describe that their test subjects were all classical musicians, and not simply picked off the street. The test material was classical, and the subjects knew how to listen to it. I think the selection of the subjects was appropriate. Classical music when well played and well heard is complex, emotive, and fascinating.
I have devoted much of my life to learning how to make and play back classical recordings that hold attention in this way, and learning how to design halls which can provide an even better experience for the majority of the audience. The papers on this site are intended to help other people understand some of the things I have learned. They are primarily preprints or power points, and are not published in standard journals. Hopefully this will change.
http://www.davidgriesinger.com/The_Physics_of_auditory_proximity.pptx
By the way Griesinger designed the only multichannel decoder that Toole likes - Lexicon MC-12 https://www.harman.com/documents/HowManyChannels_0.pdf
So far this has been just playing with words such as significant/insignificant, audible/inaudible, important/irrelevant, radical/subtle, preference etc. without specifying e.g. what is important and how much is significant for each of us, and is audible the same as perceivable in every case with speaker reproduction or not.
Also talking about "phase" is worthless without specifying difference to minimum phase i.e. excess phase/group delay, knowing that some thresholds to perceivable and audible exist.
So Juhazi's comment about phase/harmonics is perfectly valid though phase sounds insignificant without more accurate spec. e.g. excess delay causing perceivable energy smoothing to transients -> lame and non-dynamic reproduction.
I have no idea what you are talking about?
I was referring to localization at those frequencies.
When sources are (close to be) coherent obviously phase differences (delay is basically just also a phase difference) will change the amplitude.
And no it's not about laziness from my side, but rather a general laziness of people (I am NOT referring to anyone specific) to read the standard acoustics literature.
I have no idea why you start to refer localization while Juhazi was (presumably) talking about timing, and markbakk asked something about it? And we all are more or less off topic so let's shut up
Sorry for the hijack, but in fact it is quite interesting. As far as I can judge in this short time, modern science hasn't brought good enough hypotheses for the observations that mammals (including humans) are indeed able to locate low-frequency sounds to a certain extent. Griesinger simplifies things too much imho (and that anyone can check by reading modern research). And there seems to be a general understanding that masking plays an important role in localization (as well as it does in perception of frequencies and levels). The classic ITD and ILD Rayleigh model (1907!) holds somehow, but the effect of multiple sources and locations on the ability to discern and separate every one, the low frequency ones also, probably hasn't been explored enough.
A lot of the research seems to settle on frequency ranges from about 100Hz and upwards. Some sources focus on lower frequencies. When the room comes into play, things get more diffuse. Fairly recent CASA models show that subwoofers (single source) are localizable under circumstances, and within several milliseconds. I found no real clues about the phase relation of low frequency tones and their harmonics wrt to the localization of sources, so if anyone does, please give a helping hand here.
For us music listeners it is also relevant to which extent the stereo signal actually contains the needed spatial information. And there really is a tradeoff between good amplitude behavior and the spatial experience of the stereo signal in the LF domain...
[Edit]Sorry, Juhazi started the localization thing... thx for that
A lot of the research seems to settle on frequency ranges from about 100Hz and upwards. Some sources focus on lower frequencies. When the room comes into play, things get more diffuse. Fairly recent CASA models show that subwoofers (single source) are localizable under circumstances, and within several milliseconds. I found no real clues about the phase relation of low frequency tones and their harmonics wrt to the localization of sources, so if anyone does, please give a helping hand here.
For us music listeners it is also relevant to which extent the stereo signal actually contains the needed spatial information. And there really is a tradeoff between good amplitude behavior and the spatial experience of the stereo signal in the LF domain...
[Edit]Sorry, Juhazi started the localization thing... thx for that
Last edited:
Well, I wasn't talking about localisation and I believe proximity doesn't mean that either. Modern research doesn't know everything and for sure there aren't exact limits.
D&D 8c has LR4 xo at 80Hz to monopole bass that gets reflected from front wall (that's why there is specific settins). It would be interesting to study inroom wavelet decays with various settings and compare those to subjective listener opinions. I bet that has happened during development!
D&D 8c has LR4 xo at 80Hz to monopole bass that gets reflected from front wall (that's why there is specific settins). It would be interesting to study inroom wavelet decays with various settings and compare those to subjective listener opinions. I bet that has happened during development!
Bass drum and other non synthetized sounds contain quite wide bandwidth of frequencies. Try a low pass filter on 50Hz on kick drum track and checkout how it sounds Carry on
ps. It is rare to hear even a synthetized instrument with only fundamental, the lows. Harmonics are added at the mixing so that the lows are "heard" with systems that don't have subs.
Carry onps. It is rare to hear even a synthetized instrument with only fundamental, the lows. Harmonics are added at the mixing so that the lows are "heard" with systems that don't have subs.
Last edited:
Hijacking continues...
Argument for multi-sub assumes low distortion and noise so that subs do not send frequencies and extra sound which would be too easy localize after low-pass filter. Possibility to sense sound waves (/sound sources) and very high excess GD or some GD twist with IIR still remain. Advantages may still be crucial for individual no matter what I prefer.
Though typical LF instruments contain harmonics which help following melody it's still possible that bass melody is partly impossible to follow due to sharp dips and peaks in room response and pitch error they cause to transients and very short notes. There are differences how listeners notice and care about errors. Many listeners do not care a **** about accurate LF melody reproduction ~ "music itself" compared to e.g. spaciousness; feeling to be in large hall and listening echoes and space.
Argument for multi-sub assumes low distortion and noise so that subs do not send frequencies and extra sound which would be too easy localize after low-pass filter. Possibility to sense sound waves (/sound sources) and very high excess GD or some GD twist with IIR still remain. Advantages may still be crucial for individual no matter what I prefer.
Though typical LF instruments contain harmonics which help following melody it's still possible that bass melody is partly impossible to follow due to sharp dips and peaks in room response and pitch error they cause to transients and very short notes. There are differences how listeners notice and care about errors. Many listeners do not care a **** about accurate LF melody reproduction ~ "music itself" compared to e.g. spaciousness; feeling to be in large hall and listening echoes and space.
I don't think that is what he is saying. The point that subs can be localized through noise, distortion, chuffing etc. and that if not careful using subs with these problems in a multisub arrangement could easily make it less preferable than a pair of cardioid mains. I don't think that makes it any clearer which is better if both were optimized.
Bass drum and other non synthetized sounds contain quite wide bandwidth of frequencies. Try a low pass filter on 50Hz on kick drum track and checkout how it sounds
ps. It is rare to hear even a synthetized instrument with only fundamental, the lows. Harmonics are added at the mixing so that the lows are "heard" with systems that don't have subs.
Yes, and if harmonics have different GD, timbre changes. D2 of 50hz is 100Hz, and often there is LR4 xo between them
Even that could prove to be challenging, our ears have quite a high-pass filter by nature, only quite steep filtering would suffice. But for me, as we have reason to believe low frequency sound sources can be localized, but we don't yet know a satisfying explanation, it boils down to being careful with assumptions either way.Try a low pass filter on 50Hz on kick drum track and checkout how it sounds
OTOH, the question is still open about the phase relation of fundamentals and harmonics and it's perceivability. I won't make too much assumptions here, but on localization I would say: since the spatial information in the fundamental of the phantom (stereo) source remains unchanged in a full stereo setup, the localization mechanism in our ears still would function properly, notwithstanding phase shifts.
Our ears, the nervous system behind them and the presumed localization mechanism are frequency dependent. So to me, changing phase relation might well mean nothing to our ability to localize sound sources. And I do not know of research that justifies the claim that phase shifts in low to medium frequency domain themselves are perceivable. Which brings me to the assumption that group delay on low frequencies isn't that important. Even without the room equation into play. But I don't know for sure
Argument for what? I'm not selling anything here. Just saying for example that full range cardioid is valid option and could be suitable, easiest, ... the best for many no matter what doctor says.
If you were what would it be for your version of optimal for the average room?
Doctor, do you mean gedlee? I don't recall him saying that cardioid won't perform well.
As a designer of commercial speakers I try to produce profitable products. This means (at least on our local market) boxed speakers if target is to sell more than one pair in five years.
My career as a DIY designer is over so cardioids are as much history as conventional boxes.
I think that we have reached agreement about this low bass subject!
"We don't know for sure"
Back to multiway speakers...
Anyway, the thing I hate most is a bass reflex (sub)woofer! I have never heard cardioid bass, but I fully support kimmostos arguments for it. Sadly the unit must be really large (2x12" at least) so eg. 2x8" sealed might be second best. Dipole must me as big, but is much more difficult regarding placement.
Next question follows - where would you set W/M xo? Lately I have made two 3-way towers that use sealed woofers and crossed around 300Hz LR2. I really like the bass tightness and thump they give! Positioned close to front wall, which means maximal mode excitation...
D&D8c and my AINOgradients have similarity in using sealed monopole bass, AINOs up to 150Hz LR2 to dipole which means cardioid 100-200Hz... see my avatar. Actually they give pretty good bass sound but my living room is open concept with minimal modes...
"We don't know for sure"
Back to multiway speakers...
Anyway, the thing I hate most is a bass reflex (sub)woofer! I have never heard cardioid bass, but I fully support kimmostos arguments for it. Sadly the unit must be really large (2x12" at least) so eg. 2x8" sealed might be second best. Dipole must me as big, but is much more difficult regarding placement.
Next question follows - where would you set W/M xo? Lately I have made two 3-way towers that use sealed woofers and crossed around 300Hz LR2. I really like the bass tightness and thump they give! Positioned close to front wall, which means maximal mode excitation...
D&D8c and my AINOgradients have similarity in using sealed monopole bass, AINOs up to 150Hz LR2 to dipole which means cardioid 100-200Hz... see my avatar. Actually they give pretty good bass sound but my living room is open concept with minimal modes...