Thanks for your answer, @Floyd Toole. I understand that resonances show up as bumps in all the curves - and pulls the preference score down. Perhaps I wasn’t clear with my question, so let me ask in this way: Can we rule out that a loudspeaker can have a «non-smooth» response (in all the curves) where the «non-smoothness» is not related to resonances, but listeners would still award this loudspeakers a high score? Note, I have little knowledge about loudspeaker building, so maybe this is not technically possible.
-
Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Floyd Toole's Toronto AES presentation from October 28, 2025
- Thread starter RickS
- Start date
- Featured
-
- Tags
- aes audio presentation toole
Floyd Toole
Addicted to Fun and Learning
Audio Luminary
Technical Expert
Industry Insider
Forum Donor
What you describe is technically possible. What is at issue is whether an irregularity in a curve is caused by a resonance or by acoustical interference. What was illustrated in post 14 drew attention to resonances that repeat in the family of curves. This happens because the mechanisms creating the resonances radiate sound widely, so they appear in all or most measurements on and off axis, in the spatially averaged curves, and in the reflected sound field in rooms. Sounds arriving at a listener directly from the loudspeaker and reflected within the room therefore contain reinforcing information about the resonances and they are even more audible than in a non-reflective space.
Acoustical interference exists when multiple sounds from different points of origin add at a point in space. This occurs in multiple transducer multi-directional loudspeakers. This might seem also to be undesirable, but the acoustical interference takes different forms at different off-axis angles - nothing is consistently repeated. Spatially averaged curves, comprising many measurements at different off-axis angles will statistically average out most of the evidence of interference, but not the evidence of resonances. In normally reflective rooms these inconsistently non-smooth off-axis sounds arrive at listeners from many angles at different times achieving a different form of statistical "smoothing".
Conventional forward-firing cone and dome loudspeakers are simple sources and acoustical interference can be minimized - eliminating resonances is the prime objective. Multiple transducer multi-directional loudspeakers also need to be free from resonances, but will exhibit off-axis acoustical interference which is acoustically and perceptually "averaged" by the reflected sound field in normal rooms. The broadband spectral balance of all radiated sounds should be similar, though. So, the simple answer to your question is that loudspeakers designed to deliver a resonance-free flattish and smooth direct sound and timbrally similar reflected sound fields should be subjectively highly rated in terms of sound quality.
There will be significant differences in the spatial illusions resulting from such widely different designs, and these spatial effects figure strongly in subjective preferences, but that is a different topic.
This is a simplification, ignoring large-panel radiators with their own issues, for example, but I think fundamentally correct.
bennybbbx
Addicted to Fun and Learning
I completely agree!
and because we have no laser vibrometer, and i guess near all speaker developers too, maybe this is a good replace ?. Near field measure as Klippel. To create an anechoic environment, I measured the back of the Focal and the JBL speakers at a distance of 2 mm from the center of the rear panel of the enclosure. Although the JBL is at a disadvantage because the bass port is located at the back(focal on front), there are significantly fewer resonances at 140 ms, resulting in a faster decay. speakers with big heatsink can reduce resonances, but class D speaker developers should recognize that and damp their backplates more.
So i come to conclusion this caused by structural cabinet compliance and thin metal back While a parametric EQ can flatten the steady-state amplitude frequency response, it cannot stop a structurally weak cabinet from physically storing kinetic energy and bleeding it back into the room over time or in short called resonance. big cases can emit more energy
JBL at 2 ms backside measure and measure distance ~2 mm distance to backside of JBL
same with the focal
JBL at 140 ms backside measure and measure distance ~2 mm distance to backside of JBL
the focal same situation. the 2 ms level of the focal at 45 hz was ~73.2 db. 140 ms later this decay only to 70 db. the jbl have at 2 ms level of 75 db . 140 ms later this decay to 61 db.
Then can say decay level of
focal in 140 ms = ~3 db (exact between 3.2 and 3.3 db)
JBL in 140 ms = 14 db.
When say this can not hear, because of room, then please also answer the question. when not hear wy should buy more expensive speakers. the FR at 0 ms i come to conclusion is not so important. look at headphones, they are mostly worser in FR as speakers, but of course much better in decay measures
Here i have backside measure of kali lp6 1 generation and focal. both have simular port contruction. I think the kali sound more worse as the focal. and can more resonances measure. but the kali cost only half of $ as the focal.
Thanks again, @Floyd Toole!
bennybbbx
Addicted to Fun and Learning
attached are the measure file REW mdat and a group delay screenshot from backside measure of kali, focal and jbl.
Why do speaker reviewers need a $100,000 laser system when a simple 2 mm nearfield microphone measurement captures exactly what matters to our ears?
A laser vibrometer measures the structural displacement of the panel itself, even if parts of the panel move out-of-phase and cancel each other out acoustically. But as listeners, we only care about the acoustic energy that is *actually radiated into the air*. If a flexing cabinet panel modulates the air, a 2 mm nearfield mic will capture that acoustic energy unmasked, long before room reflections can corrupt the signal, and because of long distance in compare to direct signal have very low level.
In fact, these very acoustic cancellations and out-of-phase panel movements are exactly what cause the wild, jagged fluctuations and sharp dips in the **Group Delay (GD)** plot. The extreme phase flips at the microphone capsule are a direct acoustic fingerprint of that chaotic panel behavior.
Furthermore, there is a major blind spot regarding room interaction and laser/accelerometer measurements:
A structurally weak, compliant loudspeaker cabinet inside a small room will be physically set into vibration *from the outside* by the immense acoustic pressure of the room's own modal resonances. If a room mode at 45 Hz builds up massive pressure, it will force the large MDF panels of a speaker like the Focal to flex.
Therefore, a laser vibrometer or a physical accelerometer mounted to the cabinet will inevitably measure and display the room's modal energy acting upon the speaker structure! The laser is not a pure "anechoic" isolation tool in a real room.
Given that a professional Polytec Scanning Laser Vibrometer setup costs anywhere from $80,000 to over $150,000, it is financially impossible for independent developers or reviewers to use them.
So a time-slice comparison (140 ms) or group delay via a 2 mm nearfield microphone measurement provide a significantly more practical, cost-effective, and audiologically relevant "quasi-anechoic" metric to expose whether a cabinet is acoustically inert or a resonant energy storage device
Why do speaker reviewers need a $100,000 laser system when a simple 2 mm nearfield microphone measurement captures exactly what matters to our ears?
A laser vibrometer measures the structural displacement of the panel itself, even if parts of the panel move out-of-phase and cancel each other out acoustically. But as listeners, we only care about the acoustic energy that is *actually radiated into the air*. If a flexing cabinet panel modulates the air, a 2 mm nearfield mic will capture that acoustic energy unmasked, long before room reflections can corrupt the signal, and because of long distance in compare to direct signal have very low level.
In fact, these very acoustic cancellations and out-of-phase panel movements are exactly what cause the wild, jagged fluctuations and sharp dips in the **Group Delay (GD)** plot. The extreme phase flips at the microphone capsule are a direct acoustic fingerprint of that chaotic panel behavior.
Furthermore, there is a major blind spot regarding room interaction and laser/accelerometer measurements:
A structurally weak, compliant loudspeaker cabinet inside a small room will be physically set into vibration *from the outside* by the immense acoustic pressure of the room's own modal resonances. If a room mode at 45 Hz builds up massive pressure, it will force the large MDF panels of a speaker like the Focal to flex.
Therefore, a laser vibrometer or a physical accelerometer mounted to the cabinet will inevitably measure and display the room's modal energy acting upon the speaker structure! The laser is not a pure "anechoic" isolation tool in a real room.
Given that a professional Polytec Scanning Laser Vibrometer setup costs anywhere from $80,000 to over $150,000, it is financially impossible for independent developers or reviewers to use them.
So a time-slice comparison (140 ms) or group delay via a 2 mm nearfield microphone measurement provide a significantly more practical, cost-effective, and audiologically relevant "quasi-anechoic" metric to expose whether a cabinet is acoustically inert or a resonant energy storage device
Floyd Toole
Addicted to Fun and Learning
Audio Luminary
Technical Expert
Industry Insider
Forum Donor
It is the speaker designers who make use of expensive scanning laser interferometers. With it they can see which panel vibrations effectively radiate sound and where structural changes or internal bracing are needed to minimize radiated sound, not waste money addressing vibrations that do not deliver resonant sounds into the listening room.
A measurement in the extreme near field of one portion of an enclosure surface, as your 2 mm, is not helpful in determining how much energy is radiated into the room. This is measured in comprehensive anechoic data, such as the spinorama. Resonances are bothersome only if they are audible and considerable evidence exists to confirm that these anechoic data correlate with what is audible, especially at frequencies above the transition frequency. At low frequencies room resonances are vastly more audible than transducer or structural resonances in competently designed loudspeakers. These need to be addressed separately. It is in the public domain literature.
Loudspeaker reviewers who are able to measure anechoic data in the far field of loudspeakers are able to anticipate whether resonances exist and the potential audibility of those resonances. They do not need $100,000 scanning laser vibrometers, an anechoic chamber or Klippel near-field scanners. With an inexpensive microphone, a laptop computer, a quiet yard or large space, and a modicum of skill and patience they can acquire usefully accurate data - but most reviewers do not.
Reviewers relying only on single-stimulus, take-it-home-and-listen-to-it methods, listening in stereo, are unlikely to be reliable indicators of anything subtle.
The "immense" pressure is inside the cabinet, not in the listening room. Any cabinet vibration, or indeed vibration of any surface in the listening room, including walls and floors, caused by sound in the room removes energy from the room sound field by converting sound energy into mechanical movement - ultimately into heat. The Focal enclosure panels are likely much smaller than many other moveable surfaces in a typical room. And don't drink from a resonant wine glass while listening.
bennybbbx
Addicted to Fun and Learning
How many grid points on the rear panel should I measure to prove that this is a distributed structural problem and leave no room for doubt about poor enclosure design?
See headphone measures(i use same microphone as for speakers Beyerdynamic MM1). Headphones are much better, and distance was on AKG 712(open headphone) ~5 mm on Sennheiser HD 300(close headphone) ~3 mm so that bass SPL is simular to speakers and Headphones.
To avoid shaking, I simply rested my hand holding the microphone against the headphone headband. but the headphones give much fewer group delay as the speakers measured in the same room, so Room is no problem for speakers measures.
The Klippel NFS measures hundreds of points, and we should easily be able to see this cabinet energy storage in a 140 ms time-slice or the group delay plot if such data were ever published by manufacturers like Focal or JBL or speaker testers with a Klippel. Both of these speakers have already been tested on ASR.
Regarding the internal pressure: I am absolutely talking about the pressure inside the cabinet. The larger and heavier cones of the Focal and Kali generate massive internal air movements and mechanical counter-forces, setting the enclosure into much stronger vibration than a small, lightweight driver can. By the time the sound waves finally hit the room walls and travel back to the microphone position, their level has already dropped. This is strictly a bass issue—the mids and highs on the backplate can easily be damped with Basotect on walls, where I don't hear any problems.
But even when I reduce the bass output of the Focal or Kali so that all speakers share the exact same frequency response (FR), the tiny speakers simply sound much better in terms of clarity.
the groupdelay in compare to headphones and speakers.
EDIT: Now all speakers and headphones have smoothing 1/12 in groupdelay. before was speakers more smoothed(1/3) as the headhphones(1/12)
The Bass decrease alot in the headphones but still much fewer group delay on headphones. thats SPL in compare to speakers . it is not align in REW to fit. I change the volume of headphones so it fit.
this is how i measure the HD 300. left is always the side with the cable
this is how i measure the AKG 712
For a closed headphone it is more important that it sit on ears, so this measure can not use to see which headphone is better in group delay. it should only show that speakers are so worse in compare to headphone in Group delay. the mdat file from the headphones for REW you can download too.
See headphone measures(i use same microphone as for speakers Beyerdynamic MM1). Headphones are much better, and distance was on AKG 712(open headphone) ~5 mm on Sennheiser HD 300(close headphone) ~3 mm so that bass SPL is simular to speakers and Headphones.
To avoid shaking, I simply rested my hand holding the microphone against the headphone headband. but the headphones give much fewer group delay as the speakers measured in the same room, so Room is no problem for speakers measures.
The Klippel NFS measures hundreds of points, and we should easily be able to see this cabinet energy storage in a 140 ms time-slice or the group delay plot if such data were ever published by manufacturers like Focal or JBL or speaker testers with a Klippel. Both of these speakers have already been tested on ASR.
Regarding the internal pressure: I am absolutely talking about the pressure inside the cabinet. The larger and heavier cones of the Focal and Kali generate massive internal air movements and mechanical counter-forces, setting the enclosure into much stronger vibration than a small, lightweight driver can. By the time the sound waves finally hit the room walls and travel back to the microphone position, their level has already dropped. This is strictly a bass issue—the mids and highs on the backplate can easily be damped with Basotect on walls, where I don't hear any problems.
But even when I reduce the bass output of the Focal or Kali so that all speakers share the exact same frequency response (FR), the tiny speakers simply sound much better in terms of clarity.
the groupdelay in compare to headphones and speakers.
EDIT: Now all speakers and headphones have smoothing 1/12 in groupdelay. before was speakers more smoothed(1/3) as the headhphones(1/12)
The Bass decrease alot in the headphones but still much fewer group delay on headphones. thats SPL in compare to speakers . it is not align in REW to fit. I change the volume of headphones so it fit.
this is how i measure the HD 300. left is always the side with the cable
this is how i measure the AKG 712
For a closed headphone it is more important that it sit on ears, so this measure can not use to see which headphone is better in group delay. it should only show that speakers are so worse in compare to headphone in Group delay. the mdat file from the headphones for REW you can download too.
Attachments
Last edited:
Floyd Toole
Addicted to Fun and Learning
Audio Luminary
Technical Expert
Industry Insider
Forum Donor
If you take the trouble to look at patterns of panel vibrations in simple shapes you will see that numerous locations must be sampled simultaneously because you need to know the direction of motion at every instant to be able to estimate summations and cancellations of radiated energy. In reality, the shapes are rarely simple, even though from the outside they may be rectangular, because of internal bracing that exists in most boxes, as well as the structural effects of transducer frames. That is why scanning laser interferometers are used by well-equipped engineers to optimize enclosure design.
However, as I have said many times before, the evidence of resonances and their audibility is in comprehensive far-field anechoic measurements. If there is a problem it is then that one goes looking for the cause, not the other way around. Heroic cabinet constructions of granite, marble or concrete make good marketing, but are excessive in the real world. The evidence seen in the time domain correlates poorly with the audibility of resonances.
Low frequency resonances in loudspeaker enclosures may exist, but in terms of audibility they are in competition with very much louder room resonances. The sounds radiated by the woofer and the enclosure are filtered through the room on the way to the listener. It is the real problem.
The fact that resonances are MORE audible when listening at distance is GREAT information, and probably something you've said before which I did not catch. THANK YOU!!
Okay I was aware of this part; I'd failed to connect the dots and appreciate the implications of listening distance. Thank you for making that connection for me!
My (sighted listening) opinion is that large rooms often sound better than small rooms, timbrally as well as spatially, and maybe the longer-lasting reflection field of a good big room plays a role? Of course the bass quality is likely to be better in a larger room, and maybe that's the dominant factor.
While I'm an advocate of relatively narrow-pattern configurations - waveguides and big midwoofers - I like to inject more reflections than normal into the playback room via a secondary source whose output arrives after some delay. To my ears, the net result is a usefully closer approach to that delicious timbral richness of a good seat in a concert hall or recital hall.
While I have not done anything remotely approaching the depth of research you have, this is consistent with my observations from involvement with multi-directional loudspeakers.
In his later years David Smith, designer of (among other things) the landmark JBL Model 4430 studio monitors, explored using a secondary source to make fairly large frequency response corrections to the reflection field, to the point of correcting the system's perceived spectral balance. He posted about this on DIY Audio. David explored perceptually minimizing peaks in the direct sound as well as perceptually filling-in dips in the direct sound via his secondary array. I can find and post links if you would like. Unfortunately he passed away before I stumbled across his posts, as I would have very much liked to converse with him on the subject.
The first time this really jumped out at me was the first time I heard what might be called a controlled-pattern dipolar loudspeaker: 90 degree pattern width, very uniform across that width, and with half of the output being the spectrally-identical rear-firing (and therefore reflection-field-only) backwave.
This is a bit different from the addition of deliberately corrective reflection-field energy that David Smith investigated, and imo the ideal would be that any additional reflection-field energy would not need to be corrective.
Agreed. Imo the spatial quality consequences of multi-directional speakers are more significant than the sound quality consequences, but that is definitely a different topic.
Last edited:
bennybbbx
Addicted to Fun and Learning
Should i use FEM structoral software Simulation to confirm ?.
EDIT: I put all in quotes what the google AI write
Is it not possible, when the room resonances and the loudspeaker enclosures resonance frequency at same frequency it give a stronger additional excite to the room and boost that frequency ?. so this is another reason that speaker reviewers show the time slice of later time, maybe 140 to see it
Last edited:
Floyd Toole
Addicted to Fun and Learning
Audio Luminary
Technical Expert
Industry Insider
Forum Donor
What happens in the extreme near field is only important if it reaches the ears in a room. That is what comprehensive anechoic measurements are able to reveal. If the "destructive drone" is sufficiently energetic to be audible, it will be visible in the spinorama curves. If not, it is just a problem if you place your ear at 2 mm from the panel. I cannot say more.
You seem to be a smart guy, so figure it out. I'm off the air. Bye.
bennybbbx
Addicted to Fun and Learning
Sorry, maybe you understand wrong, this sentence about the drone to our ears was from the google AI create. all text between " " before was from google AI. i change my previous post with html quote to make it more clear that most of this code was from google AI
additional i try out more. i have a apollo solo which contain the build in RealVerb. So i create a measure with a bass reverb to simulate a worse resonating case and look if there can the bass reverb see in spinorama. case resonances are simular to reverb, it decay over time. when ask the google AI what happen with that test, it tell
Before you waste more of other people's time with you BS, you need to understand your complete lack of knowledge in this subject area. Below are 2 introductory tutorials from Acoustics Today (the quarterly magazine of the Acoustical Society of America) on structural acoustics. If you found yourself unfamiliar with the concepts in these tutorials, then learn the subject before posting any more nonsense.
Regarding speaker cabinet vibrations that don't radiate sound and why a single point near field measurement is useless ... (figure from Part 2)
bennybbbx
Addicted to Fun and Learning
Before you waste more of other people's time with you BS, you need to understand your complete lack of knowledge in this subject area. Below are 2 introductory tutorials from Acoustics Today (the quarterly magazine of the Acoustical Society of America) on structural acoustics. If you found yourself unfamiliar with the concepts in these tutorials, then learn the subject before posting any more nonsense.
Regarding speaker cabinet vibrations that don't radiate sound and why a single point near field measurement is useless ... (figure from Part 2)
ok, but which wavelength have the wave and which length have the plate in your examples ?. seem this diagram the plate is larger than the wavelength, and such a thing can only happen if a speaker case is several meter in size for the bass
Acoustics Research Group der Penn State University
the AI give that links https://www.acs.psu.edu/drussell/publications/mdqsources.pdf
and a verify this text is inside and explain what AI told
should i do real example Simulations in FEM software with wavelength and plate length , so all is clear ?
Explain first what you are trying to simulate, what results your simulations are expected to give, and why they have any relevance.
bennybbbx
Addicted to Fun and Learning
- a 45 hz wavelength = ~7 Meter
- Case of focal is ~0,35 Meter
mean waveform is 20 times larger as focal case.
If you would like to try this out in FreeCAD or Elmer FEM, simply draw the following in your CAD software: a small box (the focal point) and later a huge hollow cube around it (your room). Define the air in between as an "Acoustic Medium" and the room walls as "Reflective Boundaries." so you can also simulate how the speaker interact with your room if you simulate your room without the speaker, then simulate speaker without room. later simulate speaker and room. for simulate hear position you can add a small object too. The AI knows how FEM work and creating of objects.
Do you disagree the links i post ?
The purpose of the FEM simulation is to directly test Dr. Toole's and your hypothesis regarding structural phase cancellations on a speaker panel in the low frequencies. **What will be simulated:**I am building a 3D CAD model of a compact monitor enclosure matching the dimensions of the Focal Alpha 65 (approx. 35 cm) with 15 mm MDF panels. I will run a frequency-dependent structural and acoustic harmonic FEM simulation across the deep bass spectrum (40 Hz, 60 Hz, 75 Hz, 92 Hz, 120 Hz, and 140 Hz).**Expected results and relevance:**According to fundamental acoustics (the $ka \ll 1$ limit for sub-wavelength structures), the simulation will prove that at these frequencies, the flat cabinet panels vibrate exclusively in their 1st fundamental bending mode. This means that across all probe locations on the panel, the mechanical displacement and phase will be completely identical—the entire plate bulges outward and inward in perfect unison. The relevance is absolute: Because the simulation will show zero out-of-phase anti-nodes on the panel surface at 48 Hz or 75 Hz, **destructive acoustic cancellations in the air are mathematically impossible.** The entire cabinet panel acts as a synchronized acoustic monopole (omnidirectional radiator). Therefore, a single-point 2 mm nearfield measurement is fully representative of this in 140 ms
OP
- Thread Starter
- #37
Recommend you move this dialog to direct messaging or create your own thread.
Will move any post(s) from this thread if it helps. Thanks!
Last edited: