After some discussion in another thread I thought it might be good to discuss what a speaker resonance is as there seems to be some confusion about the term as it relates to speaker systems. Most irregularities in speakers are loosely defined by Dr. Toole as a resonance while in the classic mechanical sense may not technically be a "resonance". The way I've understood it is that anything that causes a deviation from neutral over a wide range of angles can be considered a resonance, of course it can be from many causes but the audible effect is a speaker that sounds less neutral regardless of what we want to call it. If I'm the one confused let's discuss it and maybe myself and others can learn something.
I'm going to post a few quotes to go into more detail about how Dr. Toole classifies and clarifies a resonance in speaker systems.
In Sound Reproduction: loudspeakers and Rooms 9.6.2 he says:
In sound reproduction, resonances are to be avoided. Added resonances alter the timbral character of voices and instrument in programs; they add coloration. The task of a sound reproduction system is to accurately portray the panorama of resonances and other sounds in the original sources, not to “editorialize” by adding its own.
In measurements of loudspeakers, it is common to find evidence of resonances, the normal clues being identifiable peaks in frequency response curves. In a single frequency response curve, a peak could be evidence of a resonance, or it could be the result of acoustical interference (as in the crossover region when two transducers are active). If a peak persists in a display of curves measured at different incident angles, on- and off-axis, it is highly probable that it is evidence of a resonance and not the result of acoustical interference that would be different in measurements made at different angles.
Then some good discussion in the What Science Shows thread over at AVS, I'll post a few relevant quotes when Dr. Toole was asked to clarify what a resonance is:
To Resonate or not to Resonate, That is the Question?
Apologies to Shakespeare . . .
This discussion has drifted into an area of literal interpretations of classical definitions with some semantics thrown in. If there is a shallow hump in a frequency response, in literal terms it is a very low-Q resonance, implying a mechanical, electrical or acoustical system with a "favored" frequency range. In a physical system as complex as a loudspeaker it may sometimes be difficult to decide what is happening. Crossovers are equalizers, by any other name, that interact with transducers having inherently non-flat tendencies - the result is a combination of both electrical and mechanical elements. Equalizers can be resonators just as surely as acoustical cavities, enclosure panels and cone breakup. So a frequency response feature may be partly mechanical and partly electrical , but the end result can be that of a resonance having Q. Achieving a desirable flat on-axis sound using passive or active networks can result in non-flat off-axis behavior because transducers have frequency-dependent directivity. In a room the result is that even with flat direct sound, the early reflected and later reflected sounds may exhibit emphasis over a range of frequencies that could forgivably be interpreted as a low-Q resonance.
As discussed many times in this thread, transducers are inherently minimum-phase devices, so electrical EQ can modify the performance of mechanical resonances - a huge advantage for active loudspeakers or those for which accurate anechoic data are available.
In the crossover between a 6- to 8-inch woofer and a 1-inch tweeter, a directivity mismatch at crossover is unavoidable. Above crossover, the tweeter has much wider dispersion than the woofer, so there is an energy rise over a wide frequency range. Is this a resonance? Technically not, in the dictionary definition sense. However, there is a broad hump in radiated energy, so perceptually it may appear to be so. Figure 4.13 shows such an example where even crude room curves were adequate to recognize the energy excess in an above-crossover energy excess and attenuate it. Because wide bandwidth (low-Q) phenomena are detected at very small deviations there was a clear improvement in perceived sound quality even though medium and higher-Q "real" resonances were essentially unchanged. Addressing all of the "resonances" was not surprisingly the best.
So, don't get hung up on semantics. Deviations from a linear frequency response are all describable as "resonances" if one chooses to. Broadband trends are very low-Q, narrower trends, medium Q, and so on. Even a bass tone control is an opportunity to manipulate a "resonance" - in this case the hump that develops above the low cutoff frequency which, depending on the system design will have a Q.
Narrow dips are usually the result of destructive acoustical interference and are usually audibly innocuous because they change with direction/position. Broader dips can be interpreted as anti-resonances if one chooses to, whether there is an associated frequency selective absorption process or not. Mostly not.
All fodder for more discussion
If we're being pedantic, how is this not a resonance? An increase of output over a frequency range (wide or not) seems like it would be the dictionary definition of a resonance?
Floyd Toole said:
I guess it depends on the dictionary. In my engineering schooling resonances were associated with mechanical, acoustical or electrical systems that by virtue of mass, compliance and damping - in the case of mechanical resonances or their analogs for the others - combined to favor specific frequencies by differing amounts.
The effect of these in loudspeakers is an emphasized output over a range of frequencies, so feel free to call anything that has that effect a resonance. The perceptual system will never know nor care
EDIT: The reason it really doesn't matter is that humans do not hear the ringing, we hear the spectral hump. At low Q there really is no significant ringing in any event.
Floyd Toole said:
I have not been adequately thorough in my explanations, obviously, because there can be fluctuations in frequency response on a single measurement axis caused by acoustical interference. These are audibly innocuous because they change with mic location and spatially average out. This is the reason why we focus most of our attention on the listening window data because it is sufficiently spatially averaged to attenuate evidence of interference, so that one has a chance of identifying those bumps attributable to resonances, which are the dominant sources of coloration in loudspeakers. Those bumps that persist through the increasing spatial averages of "early reflections" and ultimately "sound power" are unquestionably resonances having perceptual consequences. Whether they are solely attributable to the "classic" mass/spring kind of resonance does not matter. As I have said all loudspeakers have some amount of built in equalization - the crossover network, which often incorporates narrow band filters to shape the frequency response. Active digital networks are much more capable than the passive ones. Then there is frequency-dependent transducer directivity that modifies the off-axis radiated energy vs frequency compared to direct sound. The audible consequences of this depends significantly on the listening setup.
All of this and more is in the 3rd edition. It really is not complicated.
I'm going to post a few quotes to go into more detail about how Dr. Toole classifies and clarifies a resonance in speaker systems.
In Sound Reproduction: loudspeakers and Rooms 9.6.2 he says:
In sound reproduction, resonances are to be avoided. Added resonances alter the timbral character of voices and instrument in programs; they add coloration. The task of a sound reproduction system is to accurately portray the panorama of resonances and other sounds in the original sources, not to “editorialize” by adding its own.
In measurements of loudspeakers, it is common to find evidence of resonances, the normal clues being identifiable peaks in frequency response curves. In a single frequency response curve, a peak could be evidence of a resonance, or it could be the result of acoustical interference (as in the crossover region when two transducers are active). If a peak persists in a display of curves measured at different incident angles, on- and off-axis, it is highly probable that it is evidence of a resonance and not the result of acoustical interference that would be different in measurements made at different angles.
Then some good discussion in the What Science Shows thread over at AVS, I'll post a few relevant quotes when Dr. Toole was asked to clarify what a resonance is:
To Resonate or not to Resonate, That is the Question?
Apologies to Shakespeare . . .
This discussion has drifted into an area of literal interpretations of classical definitions with some semantics thrown in. If there is a shallow hump in a frequency response, in literal terms it is a very low-Q resonance, implying a mechanical, electrical or acoustical system with a "favored" frequency range. In a physical system as complex as a loudspeaker it may sometimes be difficult to decide what is happening. Crossovers are equalizers, by any other name, that interact with transducers having inherently non-flat tendencies - the result is a combination of both electrical and mechanical elements. Equalizers can be resonators just as surely as acoustical cavities, enclosure panels and cone breakup. So a frequency response feature may be partly mechanical and partly electrical , but the end result can be that of a resonance having Q. Achieving a desirable flat on-axis sound using passive or active networks can result in non-flat off-axis behavior because transducers have frequency-dependent directivity. In a room the result is that even with flat direct sound, the early reflected and later reflected sounds may exhibit emphasis over a range of frequencies that could forgivably be interpreted as a low-Q resonance.
As discussed many times in this thread, transducers are inherently minimum-phase devices, so electrical EQ can modify the performance of mechanical resonances - a huge advantage for active loudspeakers or those for which accurate anechoic data are available.
In the crossover between a 6- to 8-inch woofer and a 1-inch tweeter, a directivity mismatch at crossover is unavoidable. Above crossover, the tweeter has much wider dispersion than the woofer, so there is an energy rise over a wide frequency range. Is this a resonance? Technically not, in the dictionary definition sense. However, there is a broad hump in radiated energy, so perceptually it may appear to be so. Figure 4.13 shows such an example where even crude room curves were adequate to recognize the energy excess in an above-crossover energy excess and attenuate it. Because wide bandwidth (low-Q) phenomena are detected at very small deviations there was a clear improvement in perceived sound quality even though medium and higher-Q "real" resonances were essentially unchanged. Addressing all of the "resonances" was not surprisingly the best.
So, don't get hung up on semantics. Deviations from a linear frequency response are all describable as "resonances" if one chooses to. Broadband trends are very low-Q, narrower trends, medium Q, and so on. Even a bass tone control is an opportunity to manipulate a "resonance" - in this case the hump that develops above the low cutoff frequency which, depending on the system design will have a Q.
Narrow dips are usually the result of destructive acoustical interference and are usually audibly innocuous because they change with direction/position. Broader dips can be interpreted as anti-resonances if one chooses to, whether there is an associated frequency selective absorption process or not. Mostly not.
All fodder for more discussion
...
In the crossover between a 6- to 8-inch woofer and a 1-inch tweeter, a directivity mismatch at crossover is unavoidable. Above crossover, the tweeter has much wider dispersion than the woofer, so there is an energy rise over a wide frequency range. Is this a resonance? Technically not, in the dictionary definition sense. ...
motrek said:If we're being pedantic, how is this not a resonance? An increase of output over a frequency range (wide or not) seems like it would be the dictionary definition of a resonance?
Floyd Toole said:
I guess it depends on the dictionary. In my engineering schooling resonances were associated with mechanical, acoustical or electrical systems that by virtue of mass, compliance and damping - in the case of mechanical resonances or their analogs for the others - combined to favor specific frequencies by differing amounts.
The effect of these in loudspeakers is an emphasized output over a range of frequencies, so feel free to call anything that has that effect a resonance. The perceptual system will never know nor care
EDIT: The reason it really doesn't matter is that humans do not hear the ringing, we hear the spectral hump. At low Q there really is no significant ringing in any event.
Does this mean if I boost frequencies via my AVR’s equalizer DSP feature, I’m adding acoustic resonances?
Yes. As I said, because loudspeaker transducers are minimum-phase devices one can use electrical parametric EQ to attenuate the mechanical resonances in transducers - using anechoic data of course. So, if you add a hump to an otherwise neutral/resonance free speaker you have added a resonance. This is why it is crucial to pay attention to what "room equalizers" are doing. If they "see" a ripple in a measured curve caused by acoustical interference of direct and reflected sound, and try to flatten it, they may be adding a resonance and degrading a good loudspeaker.Floyd Toole said:
I have not been adequately thorough in my explanations, obviously, because there can be fluctuations in frequency response on a single measurement axis caused by acoustical interference. These are audibly innocuous because they change with mic location and spatially average out. This is the reason why we focus most of our attention on the listening window data because it is sufficiently spatially averaged to attenuate evidence of interference, so that one has a chance of identifying those bumps attributable to resonances, which are the dominant sources of coloration in loudspeakers. Those bumps that persist through the increasing spatial averages of "early reflections" and ultimately "sound power" are unquestionably resonances having perceptual consequences. Whether they are solely attributable to the "classic" mass/spring kind of resonance does not matter. As I have said all loudspeakers have some amount of built in equalization - the crossover network, which often incorporates narrow band filters to shape the frequency response. Active digital networks are much more capable than the passive ones. Then there is frequency-dependent transducer directivity that modifies the off-axis radiated energy vs frequency compared to direct sound. The audible consequences of this depends significantly on the listening setup.
All of this and more is in the 3rd edition. It really is not complicated.
