Oh if only.. Actually I think it's pub time.
Anyway back to speaker stuff. On directivity of a loudspeaker and what will work best in a room, give the best illusion etc... I used to think that theoretically the ideal loudspeaker would be omni directional approximately 20-20kHz. I'm not so sure that is exactly true. Not saying it wont work well - it certainly would but that isn't the point.
If you were to take a competent musician and a speaker which replicates the directivity of that musicians instrument in the same room and have them play next to each other the physical replication of the musician would be so similar it would be very hard to tell the difference with a blind comparison. Part of this is that the speaker and player are in the same room but a larger part of this is that it is just analytically very close to the soundfield that you are trying to replicate.
Stereo of course is different and is in part illusion bases (in terms of a phantom image). So I guess to me the ideal loudspeaker would be one that could alter it's directivity in a polyphonic and extremely controlled fashion. It should also be able to control resonance in terms of turning them on and off on a dime in a polyphonic fashion. In other words it's not gonna happen anytime soon.
So all theory aside what should we do now? I think it's subjective but there are certain patterns under the surface of directivity in loudspeakers that we just haven't identified. And if we did have the theoretical speaker which could change directivity we would be able to dial in these differences and hear exactly what is going on. But we have not managed to achieve anything close to this imo. The closest thing I have heard of is Linkwitz with his omni and dipole. Maybe speaker manufacturers have done similar things but I just haven't seen the useful data that helps me as someone who wants to control and make music.
What do mastering engineers do when they can't control all the instruments and are basically unable to make all of them optimized due to the limitations of 2 tracks? I read Steve Hoffman talking about how he goes about mastering and it's basically an acceptance of this problem. I can't remember exactly what he said but basically he concentrates on the stand out instrument. He zones in on that one instrument and tries to make it sound as real as he can make it sound. Maybe some other instruments suffer but that is the compromise that has to be made possibly.
So I guess it's my thinking that the speaker which gives the best illusion of the human voice and it's directivity will be the speaker which performs the most consistently. If you can get rid of that weak link then the rest will fall into place. Of course there is more to it, I could be wrong, and YMMV.
Anyway back to speaker stuff. On directivity of a loudspeaker and what will work best in a room, give the best illusion etc... I used to think that theoretically the ideal loudspeaker would be omni directional approximately 20-20kHz. I'm not so sure that is exactly true. Not saying it wont work well - it certainly would but that isn't the point.
If you were to take a competent musician and a speaker which replicates the directivity of that musicians instrument in the same room and have them play next to each other the physical replication of the musician would be so similar it would be very hard to tell the difference with a blind comparison. Part of this is that the speaker and player are in the same room but a larger part of this is that it is just analytically very close to the soundfield that you are trying to replicate.
Stereo of course is different and is in part illusion bases (in terms of a phantom image). So I guess to me the ideal loudspeaker would be one that could alter it's directivity in a polyphonic and extremely controlled fashion. It should also be able to control resonance in terms of turning them on and off on a dime in a polyphonic fashion. In other words it's not gonna happen anytime soon.
So all theory aside what should we do now? I think it's subjective but there are certain patterns under the surface of directivity in loudspeakers that we just haven't identified. And if we did have the theoretical speaker which could change directivity we would be able to dial in these differences and hear exactly what is going on. But we have not managed to achieve anything close to this imo. The closest thing I have heard of is Linkwitz with his omni and dipole. Maybe speaker manufacturers have done similar things but I just haven't seen the useful data that helps me as someone who wants to control and make music.
What do mastering engineers do when they can't control all the instruments and are basically unable to make all of them optimized due to the limitations of 2 tracks? I read Steve Hoffman talking about how he goes about mastering and it's basically an acceptance of this problem. I can't remember exactly what he said but basically he concentrates on the stand out instrument. He zones in on that one instrument and tries to make it sound as real as he can make it sound. Maybe some other instruments suffer but that is the compromise that has to be made possibly.
So I guess it's my thinking that the speaker which gives the best illusion of the human voice and it's directivity will be the speaker which performs the most consistently. If you can get rid of that weak link then the rest will fall into place. Of course there is more to it, I could be wrong, and YMMV.
Interesting thought ...
Measurement of 3D Phoneme-Specific Radiation Patterns in Speech and Singing - 2007 Scientific Report
Examination of human voice properties is most interesting, since
there are often analogies between those properties and
our hearing since speech perception is one of the major tasks
of our hearing.
As an example i am convinced that is is advantageous, to keep
XO frequencies out of the "directional bands" found by BLAUERT.
Especially if the directivity has discontinuities around the crossover
frequency of a speaker.
The above exercise shows, that directivity of vowel frequencies
is speaker dependent to some degree - which may lead to
speculation whether those directivity patterns belong to
what we call "speaker identifying features"...
Although i was confronted with speaker specific features during
development of speech recognition systems, i had never
examined those directivity patterns systematically.
But there are patterns to be found even when averaged over many
speakers. Those may truly be worth to take into consideration, when
judging directivity patterns of loudspeakers.
Another circumstance is that spoken and sung vowels e.g. may differ
considerably in spectral distribution and differences in spectral directivity
are to be expected also ...
Last edited:
As a first common statement we can imagine the human voice to
be a directional sound source, having a mainly kardioid pattern
thereby getting more directional with frequency.
That "human voice sound source" is in a common communicative
situation pointing at the listener with its main direction of radiation.
Trivial, but interesting to keep in mind.
be a directional sound source, having a mainly kardioid pattern
thereby getting more directional with frequency.
That "human voice sound source" is in a common communicative
situation pointing at the listener with its main direction of radiation.
Trivial, but interesting to keep in mind.
Last edited:
Hmm, let's see, Loudspeakers and room as a system.
Nope, none of the last ten posts had anything to do with the title. Sure was fun weeding through all that useless banter though. I know, why don't you two take a break from this thread. Might do you some good.
To say something about DMLs that belongs to the topic:
According to this paper they have sharper wavefronts than conventional speakers and are more sensitive to room modes.
http://www.essex.ac.uk/csee/researc...ubdocs/C123 Modeling romm interaction DML.pdf
For more about these speakers see this thread:
http://www.diyaudio.com/forums/planars-exotics/133711-piezo-nxt-type-panel.html
From post 422 on you find theoretical discussions.
Speaker-Room Interaction and DML
This paper is unfortunately in german language
http://www.wvier.de/texte/NXT_Tonmeistertagung.pdf
but from pictures 11 ... 12, which show interference effects between
DML and neighboured walls, i would interpret the result in
exactly the opposite direction.
I could not find these exercise in english language, but the
picture from the experimental setting and the resulting plots
should be self explaining to some degree.
This paper is unfortunately in german language
http://www.wvier.de/texte/NXT_Tonmeistertagung.pdf
but from pictures 11 ... 12, which show interference effects between
DML and neighboured walls, i would interpret the result in
exactly the opposite direction.
I could not find these exercise in english language, but the
picture from the experimental setting and the resulting plots
should be self explaining to some degree.
taking into account the directivity of the flooder (without tilting it towards the listener) it perhaps, to some degree, can be seen as an equivalent of an expansion ceiling rather than compression
quoted from: whats a 'compression ceiling' - Gearslutz.com
best regards,
graaf
Hmm, let's see, Loudspeakers and room as a system.
Nope, none of the last ten posts had anything to do with the title. Sure was fun weeding through all that useless banter though. I know, why don't you two take a break from this thread. Might do you some good.
excuse me Officer but in my most humble opinion things like Sonab speakers, controlled directivity, room tweaks, evolution of Hegeman's designs from Lowther to Harman, DML interaction with rooms, "early reflections or not" or even compression ceilings and "ultimate Klaus Schulze Speaker" (oh well
)are strictly on topicwho knows the topic of a thread?
I say - ask the user who started it.
The topic is more or less defined in the first post, isn't it?
In case of this thread it is about various ideas of making of loudspeakers and rooms that work together as one system (as Roy Allison undestood it, the title of this thread is taken from Roy Allison) as opposed to making room-blind loudspeakers or killing the bad room to make it acoustically dead
THIS is the topic of this thread in my most humble opinion as a user who started it.
If I may have any say...
best regards,
graaf
ps.
I greatly appreciate that You didn't delete those posts as OT
Last edited:
I am also curious
c'mon Key - tell us about Your system please, be so kind!
Concerning the "sensitivity to room modes" i guess i have found the citation
from Page 6 in that paper:
"It was found that the results for the DML were more sensitive to
the number of room modes modelled than the results for the piston."
I interpret that statement as follows:
The computational ressources for modelling the soundfield of a DML
in a realistic manner are higher, as you need to take more
room modes into acount.
This does not imply a DML to interact more "pronounced" with
room modes. The statement is aiming at the computational effort
in the model solely IMO.
When comparing figures 3 and 4 you can see, that for the "880Hz" example
the sound pressure distribution in the room for 50% reflecting walls is
more evenly for DML than for the conventional piston acting speaker.
That difference is not as pronounced at the 440Hz plot, where both
systems perform similar.
So both papers do not differ in their qualitative statements concerning
sound pressure distribution in a room, when comparing
piston acting speakers and DML.
That findings correlate with my own listening impressions with DML.
---
I have to read further to understand that "sharper wavefront" issue ...
Last edited:
"
Comparing Figures 5 and 6, the directivity problems of the larger piston are revealed in a lack of
sharpness in the wavefronts, particularly off-axis. By 9 ms, the multiple reflections are evident from
the plurality of wavefronts, which by this time include returns from the floor and ceiling. The
reflections are reduced by the directivity (compare Figures 5j and 6j), but the initial wavefront is also
affected. Comparing Figures 6 and 7, it can be seen that the wavefronts from the DML are even
sharper than from the small piston, indicating that the apparent source size is very small.
"
Small a p p a r e n t sound sources correlate to what we
discussed earlier as "focus".
A DML can be both:
Focussed in the projection of virtual sound sources
AND by the same time
well-tempered in the speaker-room interaction concerning smooth
pressure distribution when taking room reflections into account.
I take this paper as a further support for my statement that a
DML is able to mate the 'analytical' and the 'ambient' world
in a way impossible to a conventional speaker.
Comparing Figures 5 and 6, the directivity problems of the larger piston are revealed in a lack of
sharpness in the wavefronts, particularly off-axis. By 9 ms, the multiple reflections are evident from
the plurality of wavefronts, which by this time include returns from the floor and ceiling. The
reflections are reduced by the directivity (compare Figures 5j and 6j), but the initial wavefront is also
affected. Comparing Figures 6 and 7, it can be seen that the wavefronts from the DML are even
sharper than from the small piston, indicating that the apparent source size is very small.
"
Small a p p a r e n t sound sources correlate to what we
discussed earlier as "focus".
A DML can be both:
Focussed in the projection of virtual sound sources
AND by the same time
well-tempered in the speaker-room interaction concerning smooth
pressure distribution when taking room reflections into account.
I take this paper as a further support for my statement that a
DML is able to mate the 'analytical' and the 'ambient' world
in a way impossible to a conventional speaker.
Thank you most gracious member. I believe the ten posts I refer to are no longer visible to the public. I removed them as they had nothing to do with the title as stated. I forgot to add that the posts had been removed. Had I done such, I believe this would have avoided the obvious confusion that the gracious member encountered. You were not one of the "two" to whom I referred.
Thank you.
now I see!
yes indeed, thank You so much
obviously confused and sincerely Yours,
graaf
ps.
I am especially grateful for Your making it clear that I was not one of the "two"
It really makes me happy!
Last edited:
"And now for something completely different ..."
LA's current and unsorted thoughts concerning speaker-room interaction.
Speaker and interacting room make up a modal system.
The frequency, where the room becomes a reverberant system with
no pronounced modal bahaviour is the schroeder frequency.
That is where the modal density becomes so high that we find
>= 3 eigenfrequencies within the bandwidth of a single mode.
This modal overlap criterion holds for a room assuming, that we can
excite all the modes equally.
If we excite the room using a speaker having a certain position
and frequency dependent radiation pattern, a subset of modes
is excited, each mode to an individual extent.
In order to keep the modal overlap high
- and so the "effective schröder frequency" of the speaker/room
system as low as possible-
the speaker at its position should excite the modes of the room as
balanced as possible and make use of as many modes as possible.
This is in the first place independent from the direct/reflected sound
ratio we want to achieve at places of listening.
It is just a quality measure for the reverberant sound beeing "colored".
A speaker at an unadvantageous position and/or radiation pattern
will degrade the speaker/room system by degrading modal overlap.
---
This reminds me to the advice, not to lock-step march over a
bridge. The bridge is a modal system too.
If we hit the resonance it is only fun up to some point...
The solution is decorrelation of phase, intensity and locality
(of cause frequency too) of hits to the bridges deck.
There will be energy in most of the bridges eigenfrequencies,
but the excitation of modes will stay balanced.
That is vividly to me what a DML does.
LA's current and unsorted thoughts concerning speaker-room interaction.
Speaker and interacting room make up a modal system.
The frequency, where the room becomes a reverberant system with
no pronounced modal bahaviour is the schroeder frequency.
That is where the modal density becomes so high that we find
>= 3 eigenfrequencies within the bandwidth of a single mode.
This modal overlap criterion holds for a room assuming, that we can
excite all the modes equally.
If we excite the room using a speaker having a certain position
and frequency dependent radiation pattern, a subset of modes
is excited, each mode to an individual extent.
In order to keep the modal overlap high
- and so the "effective schröder frequency" of the speaker/room
system as low as possible-
the speaker at its position should excite the modes of the room as
balanced as possible and make use of as many modes as possible.
This is in the first place independent from the direct/reflected sound
ratio we want to achieve at places of listening.
It is just a quality measure for the reverberant sound beeing "colored".
A speaker at an unadvantageous position and/or radiation pattern
will degrade the speaker/room system by degrading modal overlap.
---
This reminds me to the advice, not to lock-step march over a
bridge. The bridge is a modal system too.
If we hit the resonance it is only fun up to some point...
The solution is decorrelation of phase, intensity and locality
(of cause frequency too) of hits to the bridges deck.
There will be energy in most of the bridges eigenfrequencies,
but the excitation of modes will stay balanced.
That is vividly to me what a DML does.
Last edited:
How would the ideal point source behave then?
Wavefront would be sharp. And the pressure distribution?