"Things that cannot be measured"

[rdenney]

Really glad you mentioned this.

It was the standard, not just the practice, for PA and Professional Cinema since at least 1982, all the major players, JBL, included, came with polar plots if they were capable of producing them. Yet when Dr. Toole started at JBL/Harman he had a tough fight to get the consumer audio people to measure off-axis. (It's a really great story he tells about this). They were "on axis +/- 3dB" is all we need.

…

Who was it that in the late 70’s said that anyone listening to any speaker in an enclosed space is listening to the radiation pattern? I think it came out of the BAS crowd. Toole built on that but there those before him.

More in response to recent posts:

We should distinguish between empirical statistical analysis and physical analysis. Voice print correlation is the former, but most audio measurements are the latter. Saying two things are fundamentally alike despite different contexts is a matter of sophisticated statistical analysis to figure which empirically observed features are determinative of similarity and which are not. We are good at that. We do not have to devolve the samples into frequency, phase and amplitude to accomplish that—we merely (!) have to reliably recognize patterns.

That is a different question as to whether we can decide the success of audio electronic design based on those deterministic (as opposed to stochastic) measurements.

Speakers seem to be subject to nonlinear relationships with their environment to a far greater extent than electronics.

Not a new topic. I recall an article in the BAS Speaker titled “The Preamplifier Myth” that sparked quite a lot of back and forth. The argument was that merely competent preamps were essentially identical by that time, in terms of what we could hear. That was 1979 or thereabouts.

Toole’s research has been about speakers, where all sorts of nonlinear behaviors are possible. Nonlinear behavior can look chaotic and purely stochastic if we don’t know the math, which we don’t always. Microphones likewise, though the behaviors of both speakers and microphones become nonlinear when they interact with the enclosed space, it seems to me, and particularly as they approach their boundary conditions. That’s why he states all the warnings that he does. But that is no excuse by golden ears to throw away what we do know that is without question determinative.

Which brings us back to empirical testing. Proving that two devices sound different in practice is conceptually easy (DBT ABX, etc.). And even if we can’t explain why they would be different, controlled testing rigorously showing the difference is still a measurement. If we can’t do that, arguing about the immeasurability of design parameters seems to me a waste of time. And yet that’s what we see over and over from those justifying their uncontrolled listening skills as being determinative.

Rick “word of the day: determinative ” Denney

 

[Cbdb2]

It's why Frank Sinatra sounds like Frank and not Dino no matter what you hear it on.

So if I sing into a computer and it spits out vocals that sound so close to Dino or Frank you cant tell the difference yould still think we dont know how to measure timbre?

AI Vocals: The 6 Best Vocal AI Plugins and Tools in 2026

AI is changing. the vocal production game faster than ever. Here's 6 AI vocal tools that'll take your tracks to new levels.

blog.landr.com

 

[kemmler3D]

I'm still wondering what a "measurement of timbre" that we could quantitatively compare to another measurement of timbre would even be. To me it doesn't seem to fit the concept.

Similarly, we can measure color with extreme accuracy. But we're not measuring "the color", We're measuring the spectrum of light reflecting off an object. We're not putting a number on something that describes a person's experience of "blueness". We're quantifying the measurable single quantities that lead to that experience.

 

[Cbdb2]

I'm still wondering what a "measurement of timbre" that we could quantitatively compare to another measurement of timbre would even be. To me it doesn't seem to fit the concept.

Similarly, we can measure color with extreme accuracy. But we're not measuring "the color", We're measuring the spectrum of light reflecting off an object. We're not putting a number on something that describes a person's experience of "blueness". We're quantifying the measurable single quantities that lead to that experience.

If you want to reproduce the color (so it looks the same as the original to every person) you measure it, who cares what an individuals "experience of blueness" is, they could be color blind. I think we have a problem because were using two different definitions of timbre. The definition that says it depends on the person, like "the color", is not measurable and also fairly useless for advancing technology. The definition that says its signal (voice/instrument) dependent is measurable, repeatable and even recreateable, like these new audio plug ins are doing.
It would be interesting to talk to somebody whos worked on these algorithms, Syncro Arts has been working on them for 25 years.

 

[Cbdb2]

Do electric guitar amps/cabs have timbre? These have also been recreated (modeled) by every other guitar pedal manufacturer. Can you model something without measuring it?

 

[Titurel]

I'm still wondering what a "measurement of timbre" that we could quantitatively compare to another measurement of timbre would even be. To me it doesn't seem to fit the concept.

Similarly, we can measure color with extreme accuracy. But we're not measuring "the color", We're measuring the spectrum of light reflecting off an object. We're not putting a number on something that describes a person's experience of "blueness". We're quantifying the measurable single quantities that lead to that experience.

We come back to the same problem. Timbre is a quality not a quantity. and thus, by definition, not subject to measurement. Try measuring "brick" - not the dimensions or weight or tensile strength of a brick, but the fundamental "brickness" of bricks. The world is chock full of immeasurable things.

 

[Travis]

Saying it again dosnt make it true. Did you miss post 2081 or just ignoring it? Theres a list of 10 characteristics of timbre and how to directly measure them. Synthesizers have been using these to recreate instruments for 40 years.

Not ignoring it, was just trying to recall where I read that they were getting close to identifying the timbral characteristics of Stradivarius violins.

Found it.

To narrow down what the timbral characteristics of what makes a Strad. a Stad. they looked at physical properties of the instrument. Sound (frequency) measurements, of whatever kind, could not get them there. They looked at properties of wood, and even did a “hammer tap” on the bridge of the violins. (This is the same brick wall they have hit with voice printing, except they are anatomical).

From these finding they are in the process of seeing if they can now use vibration studies (frequency) to be able to identify a Strad. (In the conclusion section.

All I am trying to say is, albeit very poorly, is there are timbral characteristics of voice, string instruments, like a violin, and many other things that get reproduced through a loudspeaker that don’t currently allow for the identification of a unique instrument or voice from audio measurements.

Studies, like this one are so “we can understand the characteristics of the vibration modes and sound radiation” of a violin including a Strad.

They (acoustical scientists) have been working on Strad. question for a for a very, very long time. They keep inching forward, and they are almost there and they will be able to measure what a person identifies as the timbre of a Strad (or at least what some can).

Travis

Description of previous studies and methods from the article:

“Many studies have been conducted using computer technologies, such as modal analysis using numerical simulation1, vibration of body using CT scan and laser vibrometry2, microphone array and visualization of sound radiation3, timbre analysis, sound synthesis using neural network, etc. 4 The objectives of numerical simulation
of the vibration and sound radiation of a violin are as follows: it is difficult to take measurements experimentally while changing the parameters of the mechanical properties of wood, such as density, elasticity, thickness, shape, and arch. Violins are expensive and require several months to make. Additionally, it is almost impossible to
disassemble antique violins, such as those of Stradivari and Guarneri del Gesu. As their violins are cultural assets, they cannot be touched easily. Therefore, the advantage of using numerical simulations is that we can understand the characteristics of the vibration modes and sound radiation noninvasively and without making many violins. In this study, we describe the modal analysis and sound radiation by oscillation on a bridge of violin using the finite element method (FEM).”

They site 5 studies in the References going back 25 years.

 

[kemmler3D]

We come back to the same problem. Timbre is a quality not a quantity. and thus, by definition, not subject to measurement. Try measuring "brick" - not the dimensions or weight or tensile strength of a brick, but the fundamental "brickness" of bricks. The world is chock full of immeasurable things.

This, exactly.

Just like we can measure many aspects of the Stradivarius timbre, but not the timbre itself. It either is or isn't. We can't put a number on it as a singular subjective phenomenon any more than I can put a number on your personal experience of "red".

 

[j_j]

I'm still wondering what a "measurement of timbre" that we could quantitatively compare to another measurement of timbre would even be. To me it doesn't seem to fit the concept.

It should be able to create a difference metric, it would have to be multidimensional.

 

[Travis]

Any signal can be broken down into its even or odd frequency components, no matter the length of the signal. The more samples or longer time window you use, the more accurate the frequency resolution.

JJ, “since you are here” can you help me understand why, in some cases, the use of laser vibrometry (mechanical) is used to measure distortion? Where I’m familiar with is in connection with testing horn loaded bass/sub-woofer modules in professional cinema systems. As it was explained to me, the cone(s) in a horn loaded speaker for the bottom end move so little compared to a direct radiator (from a 1/10 to even 1/100), that the only way to get an accurate measurement is to do it physically.

I know that the Klipple NFS has this option (I put their website info on this down below) and they state “Only a laser sensor based on the triangulation principle measures the dc component of the displacement reliably, which is a very valuable symptom for loudspeaker nonlinearities.”

Can you explain what this means in terms of measuring distortion (whether it’s AM, FM, IM, harmonic) and what that tells us, if anything compared to the acoustic measurements?

Thanks in advance,

Travis

LASER DISPLACEMENT SENSOR

• Dynamic measurements of displacement up to 50 kHz
• Static measurement of DC component and geometry
• High accuracy
• Different measurement ranges
• Visible red type class 2 laser
• Calibration by user possible
• Measurement of the reflected light intensity
• Good cost-performance ratio
• Ideal for loudspeaker measurements

The Distortion Analyzer equipped with a laser displacement sensor becomes a powerful vibrometer providing dc and ac components of up to 50 kHz. The laser sensor gives access to the mechanical domain, simplifies the T/S parameter measurement and is the basis for the vibration and radiation analysis. Only a laser sensor based on the triangulation principle measures the dc component of the displacement reliably, which is a very valuable symptom for loudspeaker nonlinearities. This dc capability is also the basis for scanning the geometry of a radiator.

A variety of laser sensor heads is provided to get optimal performance in the particular application.
It is recommended to use coating spray for micro-speakers and tweeters with transparent and shiny diaphragms. More information.

---

Specification​

A2 Laser Displacement Sensor

---

KLIPPEL Modules Using a Laser Sensor​

R&D System
• Distortion Measurement (DIS)
• Linear Parameter Measurement (LPM)
• Large Signal Identification (LSI)
• Material Parameter Measurement (MPM)
• Power Testing (PWT)
• Scanning Vibrometer System (SCN)
• Suspension Part Measurement (SPM)
• Transfer Function Measurement (TRF)

---

Standard Measurements Requiring a Laser Sensor​

Voice coil displacement
Scanning mechanical vibration
DC displacement – dynamic offset the voice coil
Scanning geometry
Small signal lumped parameters
Transducer nonlinearities (curve shape)
Single-valued nonlinear parameters
Rub & Buzz and impulsive distortion
Compression of fundamental components
Vibration analysis
Material parameters of loudspeaker parts
Area of radiation
Waveform and spectrum
Impulse response
Magnitude and phase response
Directional radiation characteristics
Radiation analysis
Harmonic distortion
Intermodulation distortion
Accelerated life and power testing

 

[Travis]

This, exactly.

Just like we can measure many aspects of the Stradivarius timbre, but not the timbre itself. It either is or isn't. We can't put a number on it as a singular subjective phenomenon any more than I can put a number on your personal experience of "red".

We can measure aspects of timbre for a Strad., once they figure out what the significant ones are. They are zeroing in per the article I attached in a previous post. However, it looks like part of it is going to be based on vibration (mechanical/physical) vs. sound measurements. Why would they put all of this work into this? I’m guessing it’s because Yamaha or other instrument maker with that data could can make an instrument that is acoustically identical (or some percentage or factor of identical) to a Strad. (1 sold last year for I think $11M, record was few years ago, almost $16M) and charge accordingly. If they know what to measure, and can make it, they can repeat the blind studies and the subjects won’t be able to pick one from the other (unlike the prior studies where educated audience listeners (not players) were able to identify a difference.

Then there are several steps to carry that over to loudspeaker measurements (if they even can) and if they can, will it matter in terms of improving accurate sound reproduction as you and others have mentioned is the goal.

 

[IPunchCholla]

Wasn’t there a recent DBT ABX that showed people couldn’t tell the difference between strats and something much cheaper and when there was a preference for the cheaper? Shouldn’t those studies have started with evidence they have a unique timbre? Talk about begging the question.

 

[Travis]

Who was it that in the late 70’s said that anyone listening to any speaker in an enclosed space is listening to the radiation pattern? I think it came out of the BAS crowd. Toole built on that but there those before him.

Klipsch did a paper on it for the Acoustic Society journal in the 70s. He adopted it as one of his four core design principles, along with distortion, FR, and even ranked it in importance. It was a subject of debate between Holman and others in the cinema sound realm in the 70s. (I can’t remember who took an opposite view who was also a luminary in that field. But Holman didn’t publish on it I don’t think, it has come out in multiple interviews.

He kept his “breakthroughs” close to his chest. You probably remember that the first certified cinema system went to JBL, and only JBL. Holman provided a sealed processor box, no knobs or adjustments, that had the crossover frequencies, slopes, time delay, etc. The DI spec, low end Frequency minimum, and sound output of the bass came out of that.

We should distinguish between empirical statistical analysis and physical analysis. Voice print correlation is the former, but most audio measurements are the latter. Saying two things are fundamentally alike despite different contexts is a matter of sophisticated statistical analysis to figure which empirically observed features are determinative of similarity and which are not. We are good at that. We do not have to devolve the samples into frequency, phase and amplitude to accomplish that—we merely (!) have to reliably recognize patterns.

I get and agree with all of that, I wasn’t using terms of art correctly throughout. One question thought is what is proper way to refer to something that is measured through a microphone and then processed vs. something that is measured with a laser vibrometer? (Mechanical?).

You should know with things in forensic “science” there is big money to be made, probably 1,000 times more than in audio, and being able to develop a device that anyone can use like a Intoxilyizer breath testing “instrument” or one that can produce a voice print “that is a 99% match” with a push of a button is big business. Sometimes science, empirical analysis, gets left at the door. In the case of breath testing devices, the “software” that does the analysis, unlike other branch’s of science, is proprietary and not available to test, challenge. Because of the difference in physical and empirical analysis, and the baggage voice prints, have I switched to violins and posted the paper that prompted some of this in my mind.

They reference a paper on the violin study about previous testing that was done and they mention timbral analysis. I’m going to get that paper and see what exactly they measured to analyze timbre. I’m sure it is going to confirm what others have said about what we can measure that relates to timbre. That analysis confirmed that Strads and other violins made during that timeframe in Verona have a different timbre IIRC.

 

[Travis]

Wasn’t there a recent DBT ABX that showed people couldn’t tell the difference between strats and something much cheaper and when there was a preference for the cheaper? Shouldn’t those studies have started with evidence they have a unique timbre? Talk about begging the question.

Edit: the one I am referring to was pretty famous and I was to say was between 5 and 10 years ago.

Wasn’t ABX, it was concert violinists in a darkened hotel room, wearing dark sunglasses, playing a Strad. or a “cheep” $10,000 violin, and a slight majority said they had a preference for the latter. The headlines and articles about the study conveyed what you mentioned. You had to pull up the study and dig into it.

But it turns out that tone/timbre wasn’t the quality they preferred, it was the sound in the performers ear which was much lower in the performers ear, which is an issue. If you play a Strad. you almost have to get musicians ear plugs to keep from going deaf very prematurely. Audiences, on the other hand, greatly preferred a Strad., blind, because they “project” out into the audience better.

You can’t level match them, and the owner of the Strad said there is no way you are handing this to someone wearing a blindfold.

No, I don’t know what the weight difference was, but supposedly a performer familiar with a Strad. said she couldn’t tell them apart. I recall there were some Strad players in the test group, some who preferred the cheap one. That was the headline.

 

[j_j]

JJ, “since you are here” can you help me understand why, in some cases, the use of laser vibrometry (mechanical) is used to measure distortion? Where I’m familiar with is in connection with testing horn loaded bass/sub-woofer modules in professional cinema systems. As it was explained to me, the cone(s) in a horn loaded speaker for the bottom end move so little compared to a direct radiator (from a 1/10 to even 1/100), that the only way to get an accurate measurement is to do it physically.

I know that the Klipple NFS has this option (I put their website info on this down below) and they state “Only a laser sensor based on the triangulation principle measures the dc component of the displacement reliably, which is a very valuable symptom for loudspeaker nonlinearities.”

Can you explain what this means in terms of measuring distortion (whether it’s AM, FM, IM, harmonic) and what that tells us, if anything compared to the acoustic measurements?

Thanks in advance,

Travis

This is a bleeping complex question, but what I believe it's saying (I'd have to see the comment in context to be sure) is that nonlinearities that are of even order will create a cone displacement from relative zero (meaning not driven) when averaged over an entire cycle of a waveform, which is to say that the driver plus the system creates a short-term (while powered with the signal) AVERAGE displacement from the rest position. (Obviously the driver has a displacement of some sort, or there would be no output, but this is observing the resulting temporary displacement ON AVERAGE due to the even order nonlinearities.

A perfectly linear system would not have any such temporary quasi-DC displacement.

I will cheerfully admit I am somewhat out of my usual territory here, and I'm applying basic math and physics.

Note, I have seen, especially in very-low-acoustic-impedance systems, mass flow due to air nonlinearity, but I don't think this is what he's referring to.

 

[Audionaut]

why, in some cases, the use of laser vibrometry (mechanical) is used to measure distortion? Where I’m familiar with is in connection with testing horn loaded bass/sub-woofer modules in professional cinema systems. As it was explained to me, the cone(s) in a horn loaded speaker for the bottom end move so little compared to a direct radiator (from a 1/10 to even 1/100), that the only way to get an accurate measurement is to do it physically.

The sound in the horn is transformed by air compression and expansion - this can create additional acoustic distortion that is not directly related to the cone movement.

In order to understand which part of the distortion comes from the driver itself (mechanical/electromagnetic) and which from the horn path, the cone movement must be considered separately.

 

[NTK]

...
The Distortion Analyzer equipped with a laser displacement sensor becomes a powerful vibrometer providing dc and ac components of up to 50 kHz. The laser sensor gives access to the mechanical domain, simplifies the T/S parameter measurement and is the basis for the vibration and radiation analysis. Only a laser sensor based on the triangulation principle measures the dc component of the displacement reliably, which is a very valuable symptom for loudspeaker nonlinearities. This dc capability is also the basis for scanning the geometry of a radiator.
...

You can find more information on the use of the laser displacement sensor for distortion measurements/characterization in Klippel's online training page. Download the training handout in step 1.

Klippel Online Training

 

[j_j]

The sound in the horn is transformed by air compression and expansion - this can create additional acoustic distortion that is not directly related to the cone movement.

In order to understand which part of the distortion comes from the driver itself (mechanical/electromagnetic) and which from the horn path, the cone movement must be considered separately.

Yes, in high frequency horns, the pressure in a horn at full power can be quite surprising, and yes, you can see pseudo-dc shifts in the diaphragm due to that kind of nonlinearity. Often this is related to causing some level of mass flow in the thinnest part of the horn throat, or sometimes in the motor structure. My understanding of the question about Klippel was that we were talking about large drivers for low frequencies.

 

[Holmz]

Currently, it (direct measurment of timbre) has nothing to do with measuring loud speakers because we currently can't measure it.

Huh?

Timbre is indeed well known, it’s how a newborn knows its mother’s voice (actually before they are born). But we cannot directly measure it because we don’t know all of its characteristics, yet. Four or five have been identified since the 60s,

At one time, it was thought that movie audiences preferred a very limited frequency response. Then Harry Olson did his famous experiments and figured out that people preferred a limited frequency range when the reproduction chain was filled with distortion. In blind testing, he determined that people actually preferred full range reproduction if it was distortion-free. Then a lot of work was put into measuring the distortion of loudspeakers. That was the 40s and 50s.

My guess (that's all it is) is that when we have microphones and analyzers sophisticated enough

They probably are now.

… to distinguish human voice (voice print) or one violin from another (timbre changes in an instrument over time), those measurements and analysis could probably be adapted to determining speaker preference testing in the way that frequency response and directivity, distortion are now.

OK “I think BS is what I am knee jerking towards.
I am pretty sure that Floyd was doing unsighted testing.
When it is sighted then people pick all sorts of things, but it is probably hard to measure psychology things.

In other words, we are currently at 86% correlation now, as per Dr. Floyd we still have a ways to go. Will a valid direct measurement of timbre get us closer to 100%.

What I am NOT saying is that because we cannot measure timbre that it renders what we currently have of little or no value (are measurements everything or nothing). As I said from the beginning, does that fact that we cannot measure timbre directly yet significant in audio/speaker measurements. "Probably not" is what I said.

If we could measure it, to a high enough resolution so that the output can say that the opera singer in this waveform is Joan Sutherland, the guitar is a pre-war Martin, the piano is a German Steinway. Humans (some) can identify all of those things right now blind. Or is it something that most speakers are capable of reproducing good enough so it's not something that is going to help refine preference testing?

 

[valerianf]

Going back to the title of this thread, we can take the example of the sensitivity for sibilance regarding an amplifier (or AVR).
Not all amplifiers are sounding the same.
Is there a measurement that quantify the propension to sibilance for the electronics?
I have not eard about one.

I just got a Onkyo RZ30 and it is an awful amplifier regarding the sensitivity to sibilance.
A measurement resulting in a sibilance index would be welcome.