Well - basically the first wavefront is used by our hearing to determine direction. So therefore it is important that this is transient-accurate.
All reverberant sound together with the direct sound is used to determine timbre.
Real-world fullrangers don't have constant directivity. Therefore not all frequencies are reflected by the same boundaries and obstacles which further causes non-consistent temporal response off axis.
Even when reflected sound isn't taken into consideration the off-axis response is different from the on-axis response and therefore leading to different temporal response.
Regards
Charles
All reverberant sound together with the direct sound is used to determine timbre.
Real-world fullrangers don't have constant directivity. Therefore not all frequencies are reflected by the same boundaries and obstacles which further causes non-consistent temporal response off axis.
Even when reflected sound isn't taken into consideration the off-axis response is different from the on-axis response and therefore leading to different temporal response.
Regards
Charles
phase_accurate said:
absolutely true - in an anechoic environment
in real life - in reverberant space - it is also true BUT under one important condition - that the reflected versions of the first transient/wavefront are identical to the first one in terms of their "build up structure" - their "timeline"
only under this condition they can be interpreted as reflections of the first wavefront and as such ignored (basically) and not distorting spatial hearing process
this is not the case of multiway loudspeaker even when it is "transient-accurate" on-axis
reflected versions of the transient/wavefront in case of such a speaker are obviously significantly deformed
phase_accurate said:
yes indeed
phase_accurate said:
this is true but those inconsistencies are minor in comparison to any multiway loudspeaker
perhaps even negligible
why?
because directivity problems of fullrange drivers occur in the higher part of the spectrum
there is too little energy in those high frequencies in most musical sounds for a frequency response irregularity to cause significant wavefront shape change
of course such a situation might happen but not often
in real music the wavefront shape is determined by lower frequencies than those highs
just a hypothesis of course
phase_accurate said:
but we can hear "off-axis" sound only as reflections
and let me restate this: it seems to me that not all frequency response irregularities lead to significant (audible) temporal response irregularities by which I understand significant (audible)transient/wavefront shape change
what do You think?
best!
graaf
phase_accurate said:
Hi there,
i agree mostly with what was said, and graaf 's answer picks
up interesting points.
Let's make a thought experiment:
Setting 1: A speaker with constant directivity - say
omnidirectional or cylindrical radiaton pattern - is heard
in a normal living room. The speaker is assumed to be flawless.
Setting 2: The direct sound heard is the same as in setting 1.
But the reverberant part is replaced by the sound a cheap
ghetto blaster would create in the same room.
Where "cheap ghetto blaster", stands for a sound source with
patchy FR, rough discontinuities in phase response and radiation
angle.
How would both settings compare subjectively ?
My opinion: When listening to solo instruments we were maybe
able to accept setting 2, but we still would prefer setting 1.
When listening to complex (orchestral) material we would prefer
setting 1 even more, no chance for setting 2.
What is setting 2 ? It is the usual setting when listening to a
fullranger in a box, which is placed in a living room ...
Solo instruments and vocalists may sound perfect.
If You listen to complex material you'd rather run away and
sometimes prefer a multiway speaker with more constant
radiation angle.
With complex material it is not possible anymore to assign
inconsistencies between direct and reverberant sound to
the (virtual) sound source/living room interaction.
With complex material (and with greater distance to the
speakers), we need to have a stable and believable
bunch of sonic copies of the original sound in the living room,
which our hearing system is able to account for.
With some time listening and adjusting our hearing to the
living room, we are able to suppress the reverberant part,
which is realised as "just copies". But it works only if the
copies are high quality copies.
Does not work with a speaker sounding like a cheap ghetto
blaster off axis ...
btw. i am a friend of high directivity speakers. But even then
reverberant sound quality must meet direct sound quality ...
I think Sigfried Linkwitz said something similar ...
This is why i introduced backward tweets to my dipole
fullrange design.
I know why.
I can switch between settings similar to setting 1
and setting 2 easily.
Especially, when listening at higher distances to the speakers,
setting 1 is far superior.
www.dipol-audio.de
Kind regards
graaf
Basically I don't oppose to what you said. But it might be difficult to assign each effect its proper importance.
Just keep in mind that SPL, nonlinear distortion and bandwidth are also important regarding accuracy of reproduction. In this respect a generously dimensioned three-way definitely has its advantages over a single fullranger. If it is transient accurate on axis it is still more accurate IMO than the big masses that aren't transient accurate in any direction.
And don't forget that the low-end extension plays a great role regarding group-delay distortion.
Here you can see the step response of a provisional transient-perfect setup of my two-way Manger system. You see the floor-bounce at 4 ms approx, which has also an almost ideal shape and this is 30 degrees off-axis ! The crossover frequency is 300 Hz where the wavelength is slightly more than one meter. The drivers are spaced quite closely in relation to this. This is one very important point IMO in order to achieve smooth off-axis -amplitude response and -time accuracy.
http://www.diyaudio.com/forums/showthread.php?postid=1139711#post1139711
Regards
Charles
Basically I don't oppose to what you said. But it might be difficult to assign each effect its proper importance.
Just keep in mind that SPL, nonlinear distortion and bandwidth are also important regarding accuracy of reproduction. In this respect a generously dimensioned three-way definitely has its advantages over a single fullranger. If it is transient accurate on axis it is still more accurate IMO than the big masses that aren't transient accurate in any direction.
And don't forget that the low-end extension plays a great role regarding group-delay distortion.
Here you can see the step response of a provisional transient-perfect setup of my two-way Manger system. You see the floor-bounce at 4 ms approx, which has also an almost ideal shape and this is 30 degrees off-axis ! The crossover frequency is 300 Hz where the wavelength is slightly more than one meter. The drivers are spaced quite closely in relation to this. This is one very important point IMO in order to achieve smooth off-axis -amplitude response and -time accuracy.
http://www.diyaudio.com/forums/showthread.php?postid=1139711#post1139711
Regards
Charles
phase_accurate said:
I agree, it is very difficult
and final answers depend also on expectations
phase_accurate said:
yes indeed
but frequency response irregularities can be easily corrected
the same cannot be said about distortions in time domain
question of max SPL capability - of what one needs and of what one gets - is very complicated
we had very interesting discussion with Gedlee in a thread He started, perhaps You can join and add Your point of view
as to nonlinear distortions - I really don't know what to think of them and of their importance
what I am quite sure is that their audibility is very frequency dependent and that the graph of hearing sensitivity to them is certainly U-shaped
at the bottom of this "U" a good fullranger can have no higher nonlinear distortions than good multiway, take Jordan JX92s as an example: http://www.hifisound.de/oxid/out/oxbaseshop/html/0/test_pdf/EJJ-1110858.pdf
on the other hand Manger driver has relatively high nonlinear distortions: http://www.manger.cz/UserFiles/File/msw-harmonic-distortion.pdf
these are very high distortions in comparison with a good multiway but is it an audible problem? is Manger inferior in the result? I don't know
as to bandwidth - as I see it lower limit is not a problem as such but of max SPL limits - after all a headphone can start from 5 Hz
problem of higher limit is first of all very subjective but it is really a matter for separate discussion
phase_accurate said:
yes, therefore I personally avoid resonant boxes, bass reflex and so on
but who knows if and how audible is on musical samples non resonant group delay at 40 or 60 Hz?
I don't know, perhaps guys like Blauert know
phase_accurate said:
very nice
my apologies go to chuck55!
for most of discussion in this post is OT
the question of time coherence has already been answered
the question whether this time coherence is important or not is of course competely different question
and final answer depends on expectations
IMHO there is a question of accuracy of sound reproduction and quite separate question of realism
some speakers can be more accurate and less realistic and other less accurate but more realistic
let me use visual analogy - a drawing can be very accurate but it is still less realistic then a photograph which can be at he same time less accurate (slightly wrong colors, slightly out of focus and so on)
giving priority to fidelity in time domain for me means giving priority to realism even at the expense of overall accuracy
like in case of this observations on Time Domain Yoshii 9 fullrange omni speakers:
see: http://www.accessmylibrary.com/coms2/summary_0286-12828690_ITM
I am aware of limitations of such a speaker but I believe that this approach can be perfected just like traditional multiway approach concentrated on frequency domain
once again - apologies to chuck55!
best!
graaf
Late to the party...
Imo, the main attribute of "full range drivers" (be they cone or ESL) that people like has nothing to do with phase at all. Rather imo it is because the harmonic structure (the tonal aspect of the sound) is consistent from low to high.
In a speaker where one switches between drivers (crossovers) if you consider where the fundamental is and what happens when the fundamental moves from say a large woofer to a smaller midrange (or worse to a tweeter) it becomes self evident that the harmonic content changes. This is audible even with the best multi-way speakers.
So, in this regard a "bad" full range speaker sounds to the ear to be more "correct" than most multiway speakers.
It is worse when the xover points are in the <300 to >3000Hz. range, the wider that range is made, the less of this effect is audible.
To make the idea clear, imagine if a concert pianist played the left hand with a full size grand piano, and the right hand with an upright... or worse a toy piano!
That's my view on it...
_-_-bear
Imo, the main attribute of "full range drivers" (be they cone or ESL) that people like has nothing to do with phase at all. Rather imo it is because the harmonic structure (the tonal aspect of the sound) is consistent from low to high.
In a speaker where one switches between drivers (crossovers) if you consider where the fundamental is and what happens when the fundamental moves from say a large woofer to a smaller midrange (or worse to a tweeter) it becomes self evident that the harmonic content changes. This is audible even with the best multi-way speakers.
So, in this regard a "bad" full range speaker sounds to the ear to be more "correct" than most multiway speakers.
It is worse when the xover points are in the <300 to >3000Hz. range, the wider that range is made, the less of this effect is audible.
To make the idea clear, imagine if a concert pianist played the left hand with a full size grand piano, and the right hand with an upright... or worse a toy piano!
That's my view on it...
_-_-bear
I am new here but I have to say that whole phase problem is a little bit misunderstood. Phase plot informs us about phase diffrence betwen input signal ( in this case electric current) and output signal which is sound of course. There is no way that any object have zero phase in whole frequency range. Also in my opinion phase plot don't give us any information about speaker.
but why is it consistent? IMHO it is a question of preserving of the sound time line and waveforms' shapes
in the end harmonic structure is about waveform shape
what is important is time coherency which for me is preserving time line of reproduced natural sound source in the direct sound and in early reflections
I would say that there is too little energy in those high frequencies in most musical sounds for a phase shift to cause any significant - audibly objectionable - wavefront shape change
fortunately "zero phase in whole frequency range" is not needed, we just need as much time coherency as is neccesary to avoid audible deviations from the time line of reproduced natural sound source (transient and waveform shape deformations)
I think that when we can see significant deviations on the phase plot at least we can say that the speaker is not time coherent and as such imperfect
that in spite of that fact it may nevertheless sound good to some ears is entirely different question of course
best,
graaf
Last edited:
I have defined the term already above:
and I concur
yes indeed, single driver IS practically time coherentin a multi-way speaker phase plot shows indirectly time relationships between outputs of individual drivers and these are those time relationships that really matter because the degree of time coherency of the loudspeaker (how much it preserves or distorts the transient and waveform shapes) depends on them
best,
graaf
here is some interesting audio-related science that is of most interest for single driver affcionados:
"Sound-Source Recognition: A Theory and Computational Model": http://sound.media.mit.edu/Papers/kdm-phdthesis.pdf
here we have literally an answer to the question: what makes a loudspeaker a HiFi loudspeaker?
somewhere earlier in this thread I posted link to writings on time domain sound reproduction theory by Hiroyuki Yoshii of Onkyo/Time Domain/Fujitsu Ten fame (writings published in 1983)
also John Watkinson (i.a. in His article "Putting the Science Back into Loudspeakers") insists on importance of loudspeakers' performance in time domain, that is something commonly neglected in the industry
latest research in sound source recognition shows that indeed time domain is at least equally important for human auditory system as frequency domain
there cannot be HiFi quality - in the sense of creating realistic experience (when the listener hears virtual sound source and thinks it is real) - when the loudspeakers err in time domain - the transient is partitioned and waveform deformed
as Yoshii-san put it:
why do most musicians laugh at HiFi and audiophilia? especially classically trained musicians?
as Yascha Heifetz put it many years ago - "high phooey and hystereo"
IMHO the problem is that "HiFi" loudspeaker design is fundamentally flawed, it is unscientific
it is based on false generalizations inherited from nineteenth century science that hearing only "analyzes sound waves in terms of sinusoids - a Fourier spectrum"
HiFi loudspeaker should be time coherent - no transient split/deformation - one single clear step on step response measurement
and not only on-axis but omnidirectional, to make inevitable early reflections coherent with the direct sound
here is what Watkinson’s writes in the topic of passive crossovers in loudspeakers:
best,
graaf
"Sound-Source Recognition: A Theory and Computational Model": http://sound.media.mit.edu/Papers/kdm-phdthesis.pdf
here we have literally an answer to the question: what makes a loudspeaker a HiFi loudspeaker?
somewhere earlier in this thread I posted link to writings on time domain sound reproduction theory by Hiroyuki Yoshii of Onkyo/Time Domain/Fujitsu Ten fame (writings published in 1983)
also John Watkinson (i.a. in His article "Putting the Science Back into Loudspeakers") insists on importance of loudspeakers' performance in time domain, that is something commonly neglected in the industry
latest research in sound source recognition shows that indeed time domain is at least equally important for human auditory system as frequency domain
there cannot be HiFi quality - in the sense of creating realistic experience (when the listener hears virtual sound source and thinks it is real) - when the loudspeakers err in time domain - the transient is partitioned and waveform deformed
as Yoshii-san put it:
why do most musicians laugh at HiFi and audiophilia? especially classically trained musicians?
as Yascha Heifetz put it many years ago - "high phooey and hystereo"
IMHO the problem is that "HiFi" loudspeaker design is fundamentally flawed, it is unscientific
it is based on false generalizations inherited from nineteenth century science that hearing only "analyzes sound waves in terms of sinusoids - a Fourier spectrum"
HiFi loudspeaker should be time coherent - no transient split/deformation - one single clear step on step response measurement
and not only on-axis but omnidirectional, to make inevitable early reflections coherent with the direct sound
here is what Watkinson’s writes in the topic of passive crossovers in loudspeakers:
best,
graaf
Last edited:
There's also a body of research ("the truth is out there", I'll try to find it) that shows how a subtraction type crossover is almost never satisfactory. It provides perfect complementary outputs. The problem is that the two drivers being crossed between also have to have perfect complementary outputs. Since this is rarely the case, the characteristics of the LP and HP sections of the crossover have to be tweaked to compensate. It's a lot easier to do that in separate, non-interacting filters.
I wouldn't be surprised but perhaps Watkinson's crossover own active crossovers are an example of an exception to "almost never"
after all He is THE digital guru
anyway, almost all crossovers are bad and all passive crossovers are certainly bad in principle
I would say more generally that it is the question of the degree of transient and waveform shapes' deformation
therefore not always the same crossover would produce equally audibly objectionable results - it depends on the energy spectrum of sound of a particular note played by a particular instrument
the more of the energy of a particular impulse is partitioned, the more time shift is induced to a particular transient/waveform the worse is the case
but of course - if there are any such distortions in the loudspeaker there will always be cases when they will be audibly objectionable
then if we take into account that:
- the more prominent notes of most instruments fall within 200<1000 Hz and that human voices are roughly in the range of 80<1100 Hz
- that the "equal power" frequency between low and mid+high is between 250Hz and 350Hz,
- that most of the acoustic energy in music is below 8 kHz...
...we can see that "300<xo<3000" is worst but "100<xo<300<3000<xo<5000" is only a bit better and that all that is really mandatory for high fidelity is time coherent reproduction of 80<9000 Hz which is close to the original requirement for High Fidelity given by Hartley (inventor of the term) in 1927 which was "perfect reproduction between the limits of 32 and 9000 cycles"
for bass notes I suppose that the transient is composed mostly of overtones and other higher frequency content (presence, slap, attack of drums, all >2 kHz) and that therefore fundamental can be somewhat delayed without any audible problems
for overtones above 5000 Hz I suppose that they contribute little to the waveform shape, their energy is to small
so perhaps a crossover "xo<100<5000<xo" is generally ok under the condition that there is no significant overlap between the drivers that is the subwoofer is really silenced above 100 Hz and supertweeter is really silenced below 5000 Hz
best,
graaf
Attachments
Last edited:
importance of fidelity in time domain - time coherency - some more food for thought
identification of voices, musical instruments and all other imaginable sound sources, depends on time factors: shape of initial transient, resonances, and envelope variations
as the say: "timbre is not static"
o
there is not only initial transient, voices and sounds of musical instruments are wholly transient in character
frequency as such (time factors taken apart) is irrelevant, tells nothing in terms of identification, literally:
see: ASA 149th Meeting Lay Language Papers Microthythmic characteristics of musical instrument initial transients
what is individual and as such identifiable depends wholly on time factors
You should also remember that identification process - from the first wave arrival at the ears to the human being become aware of a particular known, identified sound source - is not simple and takes substantial amount of time:
add to this that the direction of sound is established in the first millisecond
250 ms from the first wave arrival at the ears to the initial phase of a sound source identification (human being becoming aware of a particular, identified sound source)
250 ms is
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 and of human hearing physiology it is quite a long time
time factors are critical for hearing and as such for sound reproduction as well
from the perspective of information theory steady state signals (such as sine waves) used by audio engineers to test quality of sound reproduction equipment are useless, completely irrelevant
as steady state signal carries no information
and hearing is about processing information
BTW this is core argument of John Watkinson in His article about "putting the science into loudspeakers"
then spatial reproduction - imaging etc.
"Spatial Hearing with Simultaneous Sound Sources: A Psychophysical Investigation":
http://ses.library.usyd.edu.au/bitst...24302whole.pdf
"everything You ever wanted to know about sound source localization and imaging"
time factors - transient timeline preservation - are crucial
"integration" of the reflected sound to the experienced "complete sound" coming from a certain direction requires particular relation of the reflected sound to the direct sound in the sense of preserving certain common characteristic
the point is that it is not the "frequency content"
WHY?
simply because it is NOT YET "established" IN physical REALITY!
it takes some time for the real tone from a voice or a musical instrument to build up and sound out as defined sound event with particular spectral (frequency) content
"Shortest possible length of a spoken English consonant (voiced stop consonants)" is 30 ms
"Fastest perceptual musical separation possible" and also "the time needed to cortically process musical elements" is 100 ms
"Shortest vowel length in normal speech" is 200 ms
see:Music 839D: Rhythm
moreover:
so the "direct sound (first wave front)" is not even analyzed by the brain in terms of frequency content because the frequency content is not yet established in the sound source itself! There are only "wild spectral fluctuations called the initial transient"
what can possibly serve as a common characteristic enabling the brain to compare, analyse and integrate reflected waves as relating to a first wave can be perhaps the "initial waveform", the shape of the first "transient attack"
perhaps this is not very reliable mechanism and can be confusing so the brain takes more samples before even the sense of direction is established?
we all know that in real life it takes some time and concentration to point the direction of sound source in darkness or with eyes closed
but in real life also is very important that:
"Fastest perceptual musical separation possible" and also "the time needed to cortically process musical elements" is 100 ms
so the brain can take many samples before the cortical process leading to conscious experience really starts
and as it has been noted, for us to become fully aware of all this around 250 ms is needed!
this can be more than RT60 of a typical furnished living room (typical listening room) which is around 0.2<0.5 ms
it means that a short sound can practically stop (the sound pressure can be -60 dB in relation to the initial wave sound pressure) BEFORE we become aware of what it is!
how about that!
the timeline of the process of hearing speech and music in a typical living room seems to be as follows:
1) the brain starts to collect all the data that might be relevant immediately (<<1 ms) after the arrival of initial wavefront at the ear (sensing fluctuation of sound pressure)
2) then the sound event – a tone from a voice or instrument – gradually builds up in a process of fighting inertia, a process that can last even 100 ms (or more?), the brain continues to collect all data, comparing and selecting what seems to be relevant. The selection is based not only on evolutionary formed physiology of the sense of hearing but also on the particular person previous experience
3) then the short sound can end around 200 ms that is before we become aware of what it is at around 250 ms after the start of the whole process
what is important from the perspective of a single driver and its alleged weaknesses is that the "direct sound" is indeed VERY relevant BUT "frequency content" of "direct sound" corresponding to "frequency response on axis" is COMPETELY irrelevant
WHY?
because there is no such thing really
the sense of "where" is established well before the sense of "what"
the so called "localization phase" of the hearing process ends before first millisecond from the initial transient arrival at the ear - this is consequence of binaural hearing and of typical size of human head (distance between the ears)
so the sense of "direct sound" as a basis for comparison so that reflections can be compared and selected for "integration to one perceived sound event" is formed < 1 ms that is when we even cannot talk about any defined frequency content of a physical sound source
there are only "wild spectral fluctuations called the initial transient"
in my hypothesis what can serve as a basis for comparison is something that perhaps can be called "initial transient wave attack curve shape", or "the shape of transient impulse rise curve"
which is the very thing corrupted by every crossover filter in multiway loudspeaker
time response (as can be seen in step response measurement) off axis of a multiway loudspeaker significantly differs from its time response on axis
then we have the answer why reflections in the case of multiway loudspeakers are very often detrimental to quality of the spatial reproduction of sound
in case of multi-way speaker the precedence effect and "integration" of reflected sound cannot work properly because the reflected sound wave is significantly different from the first one
ok, that was the last chunk of thoughts from me
excuse me some repetitions but it is a sort of compilation of older posts from various other threads
best,
graaf
identification of voices, musical instruments and all other imaginable sound sources, depends on time factors: shape of initial transient, resonances, and envelope variations
as the say: "timbre is not static"
o
there is not only initial transient, voices and sounds of musical instruments are wholly transient in character
frequency as such (time factors taken apart) is irrelevant, tells nothing in terms of identification, literally:
see: ASA 149th Meeting Lay Language Papers Microthythmic characteristics of musical instrument initial transients
what is individual and as such identifiable depends wholly on time factors
You should also remember that identification process - from the first wave arrival at the ears to the human being become aware of a particular known, identified sound source - is not simple and takes substantial amount of time:
add to this that the direction of sound is established in the first millisecond
250 ms from the first wave arrival at the ears to the initial phase of a sound source identification (human being becoming aware of a particular, identified sound source)
250 ms is
see: Music 839D: Rhythm
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 and of human hearing physiology it is quite a long time
time factors are critical for hearing and as such for sound reproduction as well
from the perspective of information theory steady state signals (such as sine waves) used by audio engineers to test quality of sound reproduction equipment are useless, completely irrelevant
as steady state signal carries no information
and hearing is about processing information
BTW this is core argument of John Watkinson in His article about "putting the science into loudspeakers"
then spatial reproduction - imaging etc.
"Spatial Hearing with Simultaneous Sound Sources: A Psychophysical Investigation":
http://ses.library.usyd.edu.au/bitst...24302whole.pdf
"everything You ever wanted to know about sound source localization and imaging"
time factors - transient timeline preservation - are crucial
"integration" of the reflected sound to the experienced "complete sound" coming from a certain direction requires particular relation of the reflected sound to the direct sound in the sense of preserving certain common characteristic
the point is that it is not the "frequency content"
WHY?
simply because it is NOT YET "established" IN physical REALITY!
it takes some time for the real tone from a voice or a musical instrument to build up and sound out as defined sound event with particular spectral (frequency) content
"Shortest possible length of a spoken English consonant (voiced stop consonants)" is 30 ms
"Fastest perceptual musical separation possible" and also "the time needed to cortically process musical elements" is 100 ms
"Shortest vowel length in normal speech" is 200 ms
see:Music 839D: Rhythm
moreover:
see: Spectral Vistas
so the "direct sound (first wave front)" is not even analyzed by the brain in terms of frequency content because the frequency content is not yet established in the sound source itself! There are only "wild spectral fluctuations called the initial transient"
what can possibly serve as a common characteristic enabling the brain to compare, analyse and integrate reflected waves as relating to a first wave can be perhaps the "initial waveform", the shape of the first "transient attack"
perhaps this is not very reliable mechanism and can be confusing so the brain takes more samples before even the sense of direction is established?
we all know that in real life it takes some time and concentration to point the direction of sound source in darkness or with eyes closed
but in real life also is very important that:
"Fastest perceptual musical separation possible" and also "the time needed to cortically process musical elements" is 100 ms
so the brain can take many samples before the cortical process leading to conscious experience really starts
and as it has been noted, for us to become fully aware of all this around 250 ms is needed!
this can be more than RT60 of a typical furnished living room (typical listening room) which is around 0.2<0.5 ms
it means that a short sound can practically stop (the sound pressure can be -60 dB in relation to the initial wave sound pressure) BEFORE we become aware of what it is!
how about that!
the timeline of the process of hearing speech and music in a typical living room seems to be as follows:
1) the brain starts to collect all the data that might be relevant immediately (<<1 ms) after the arrival of initial wavefront at the ear (sensing fluctuation of sound pressure)
2) then the sound event – a tone from a voice or instrument – gradually builds up in a process of fighting inertia, a process that can last even 100 ms (or more?), the brain continues to collect all data, comparing and selecting what seems to be relevant. The selection is based not only on evolutionary formed physiology of the sense of hearing but also on the particular person previous experience
3) then the short sound can end around 200 ms that is before we become aware of what it is at around 250 ms after the start of the whole process
what is important from the perspective of a single driver and its alleged weaknesses is that the "direct sound" is indeed VERY relevant BUT "frequency content" of "direct sound" corresponding to "frequency response on axis" is COMPETELY irrelevant
WHY?
because there is no such thing really
the sense of "where" is established well before the sense of "what"
the so called "localization phase" of the hearing process ends before first millisecond from the initial transient arrival at the ear - this is consequence of binaural hearing and of typical size of human head (distance between the ears)
so the sense of "direct sound" as a basis for comparison so that reflections can be compared and selected for "integration to one perceived sound event" is formed < 1 ms that is when we even cannot talk about any defined frequency content of a physical sound source
there are only "wild spectral fluctuations called the initial transient"
in my hypothesis what can serve as a basis for comparison is something that perhaps can be called "initial transient wave attack curve shape", or "the shape of transient impulse rise curve"
which is the very thing corrupted by every crossover filter in multiway loudspeaker
time response (as can be seen in step response measurement) off axis of a multiway loudspeaker significantly differs from its time response on axis
then we have the answer why reflections in the case of multiway loudspeakers are very often detrimental to quality of the spatial reproduction of sound
in case of multi-way speaker the precedence effect and "integration" of reflected sound cannot work properly because the reflected sound wave is significantly different from the first one
ok, that was the last chunk of thoughts from me
excuse me some repetitions but it is a sort of compilation of older posts from various other threads
best,
graaf
Last edited:
Maybe that is why, trying to find the best spot for a super/support tweeter on my w5-1611sa Tangband FR's, I had the best soundstage, with the easyest to locate instruments, with the tweeter way back or facing sidewards on top of the box.
(I made a small box for the tweeter)
As soon as I put it at the front on top of the box, soundstage diminished, and instruments seemed to "shift" a little as they played different tones. There where also more "ghost" images of voices and instruments. Very hard to describe this..
It almost seems if the soundstage was more empty and easy, with "only" the instruments and voices that were supposed to be there, when the tweeter was aligned to the side.
Without the support tweeter I was missing a lot of detail in comparison to my fullrange speakers. So it seems that in spite of the tweeter only radiating indirectly, its sound does add up in our brain as details, but doesnt help at localization of instruments. its sound probably comes "to late"...
gr. Paul
(I made a small box for the tweeter)
As soon as I put it at the front on top of the box, soundstage diminished, and instruments seemed to "shift" a little as they played different tones. There where also more "ghost" images of voices and instruments. Very hard to describe this..
It almost seems if the soundstage was more empty and easy, with "only" the instruments and voices that were supposed to be there, when the tweeter was aligned to the side.
Without the support tweeter I was missing a lot of detail in comparison to my fullrange speakers. So it seems that in spite of the tweeter only radiating indirectly, its sound does add up in our brain as details, but doesnt help at localization of instruments. its sound probably comes "to late"...
gr. Paul
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.