"Things that cannot be measured"

[Duke]

I’m sure it was probably the way I paraphrased it, and not the way you read it. I have had a string of posts here lately that got viewed differently than what I had intended.

No you said it right. In my mind's eye I was visualizing on-axis and off-axis curves, but you very clearly said "DI curve". MY bad!

 

[Travis]

The frequency response graphs of a sax and trumpet playing middle C are not identical.

The spectral response envelopes, decay, etc, are different. They don’t look like test tones.

Ergo, it is measurable.

IDing them is a different question

IDing them is what you need in order to use the measurement in a meaningful way for audio.

What do you mean you can't measure timbre?

You can measure identified components of timbre (characteristics) as you have indicated. You mentioned two of them, overtones, and the time envelope (comprised of four stages (attack (which I think they determined is most important) decay, sustain and release) are two of the characteristics.

They identified 4 or 5 key characteristics by the early 60s, but it hasn’t moved much since then. When they identify enough of the other characteristics of timbre, and have microphones and equipment sensitive enough to analyze those combined characteristics, they will be able to measure timbre directly.

 

[Beave]

IDing them is what you need in order to use the measurement in a meaningful way for audio.

Why? I don't understand.

 

[Travis]

Why? I don't understand.

When it gets to the point where you can measure timbre directly, my guess is that it will transform speaker “measurements” and timbre will be incorporated in measurements in a meaningful way. You would be able to see, objectively and scientifically, that Speaker A reproduced a (fill in blank, instrument, jazz trio, female voice) by some measure of accuracy and that Speaker B reproduced that sound by some level of accuracy. If that measure of accuracy leads to a more precise determination of preference in speaker selection it will become part of the measurements.

It would need to be at that level of precision where it can ID a voice, or an instrument, to provide useful information (what coloration, and by what amount). In other words, it would have to take things beyond what we already have (frequency sweeps at various horizontal and vertical angles to predict a preference).

 

[kemmler3D]

if the timbre of music is directly measurable we are not incorporating it into measurements yet, and that would be the next step.

Not sure if I agree with that, because as you say:

It’s a very, very difficult thing to measure directly because it’s perception.

That is where I agree. Timbre is a sum of traits we use as a sonic fingerprint for voices / instruments and it only truly exists as a perception in the ear of the listener. So, I am not sure whether it makes sense to use as a metric for audio equipment.

When we do get there it will be a leap forward on measurements as great, or greater, than Heyser’s FFT paper was.

How do you see loudspeaker measurement (or any other audio measurement) improving in this scenario? Or, what do you envision being able to measure that we can't now?
I see you replied as I was writing this.

Thinking of the ML/AI tools that can pull apart musical recordings into their constituent individual instruments, they are truly impressive, they actually do what was once universally considered impossible or at least unrealistic. But I don't think they can tell us anything about speakers a Klippel can't.

It would need to be at that level of precision where it can ID a voice, or an instrument, to provide useful information (what coloration, and by what amount). In other words, it would have to take things beyond what we already have (frequency sweeps at various horizontal and vertical angles to predict a preference).

I think I get what you're talking about better from this.

It seems like you're envisioning a bridge between objective measurements and subjective descriptions, maybe? Or would it be a numerical comparison, like "Speaker A reproduces a violin with 94.2% accuracy" ? Or something else?

 

[Beave]

Now I'm more confused. Timbre is a quality of the source (the instruments, the singer's voice, etc), and has nothing to do with measuring speakers or electronics.

If the speaker (or electronics) reproduce the recorded signal with flat frequency response and low nonlinear distortion, then timbre from the original recording is preserved.

 

[Ashl777]

Now I'm more confused. Timbre is a quality of the source (the instruments, the singer's voice, etc), and has nothing to do with measuring speakers or electronics.

If the speaker (or electronics) reproduce the recorded signal with flat frequency response and low nonlinear distortion, then timbre from the original recording is preserved.

So lets say for argument sake pavarotti and bon jovi hits an exact same note. It will measure the same wont it? Will it sound the same?

 

[masterhw]

So lets say for argument sake pavarotti and bon jovi hits an exact same note. It will measure the same wont it? Will it sound the same?

no, it would not measure the same. music is virtually never pure tones, what you hear from a voice or an instrument is full of overtones and harmonics that make it recognizably (and distinctly) that instrument.
it can be a really neat exercise to sit down and actually see what’s recorded when you play a note on a piano. or a guitar. pure waves don’t exist in the real world!

 

[Beave]

So lets say for argument sake pavarotti and bon jovi hits an exact same note. It will measure the same wont it? Will it sound the same?

Of course it won't. The two captured waveforms would have the same fundamental frequency but different harmonics, different envelopes.

That's all well known - and has nothing to do with measuring speakers or electronics.

 

[j_j]

So lets say for argument sake pavarotti and bon jovi hits an exact same note. It will measure the same wont it? Will it sound the same?

Good grief, no, not even slightly. Pitch is just one of many parameters. In many ways not even the most important.

 

[kemmler3D]

It will measure the same wont it?

If you only look at the pitch, yes.

But, if you look at the waveform

or spectrogram

(opera singer)

Or spectrum

(mariah carey)

they will look pretty different. Different ways of measuring the same sound to see different aspects of it.

 

[Audionaut]

When it gets to the point where you can measure timbre directly, my guess is that it will transform speaker “measurements” and timbre will be incorporated in measurements in a meaningful way.

This is all measurable and can also be made visible in various forms of visualization.

The problem is really not measurability.

All of the Field of Audio can be measured today, but you also have to be able to interpret it.

In the field of loudspeakers, the basic technology has basically not changed since the invention of electrodynamic loudspeakers by Edward Kellogg and Chester W. Rice in 1925.
A modern BMS or Purifi driver is based on the same principles of the spring mass system.
Evolutionarily improved, but the same principle - with all its flaws and measurable limitations.

There is some research in other directions, like MBL and Manger or the Plasma Ion Tweeter, but basically the speaker in most living rooms is the same old principle.

It would therefore always be necessary to measure the living room with the specific loudspeaker in order to arrive at truly enlightening measurements.
This is being practiced more and more with amplification and DSP support.

Incidentally, this has been common practice in PA technology for a long time.

In any case, I always have to smile when someone talks about a value of at least 120 Sinad and the situation then shows the great DAC in combination with two small loudspeakers in a resonating shelf.

Because every angle of change in speaker position, every change in distance, every turn of the head, every item lying in the acoustic path, every cushion on the sofa, every wall hanging, every glass door in the wardrobe, every carpet, every room resonance and above all every single wall made of stone or a resonating wooden frame with damping filling makes a far more significant and, above all, audible and measurable difference than the difference between 90 and 120 db SINAD at an Digital Audio Converter.

Pushing the limits of what can be measured in audio technology out of the audible range makes sense, but only if the same effort is put into the fundamentals of room acoustics.

The room interacts with the loudspeaker and I would go so far as to say that the listening room is part of the loudspeaker itself.

 

[Waxx]

This is all measurable and can also be made visible in various forms of visualization.

The problem is really not measurability.

All of the Field of Audio can be measured today, but you also have to be able to interpret it.

In the field of loudspeakers, the basic technology has basically not changed since the invention of electrodynamic loudspeakers by Edward Kellogg and Chester W. Rice in 1925.
A modern BMS or Purifi driver is based on the same principles of the spring mass system.
Evolutionarily improved, but the same principle - with all its flaws and measurable limitations.

There is some research in other directions, like MBL and Manger or the Plasma Ion Tweeter, but basically the speaker in most living rooms is the same old principle.

It would therefore always be necessary to measure the living room with the specific loudspeaker in order to arrive at truly enlightening measurements.
This is being practiced more and more with amplification and DSP support.

Incidentally, this has been common practice in PA technology for a long time.

In any case, I always have to smile when someone talks about a value of at least 120 Sinad and the situation then shows the great DAC in combination with two small loudspeakers in a resonating shelf.

Because every angle of change in speaker position, every change in distance, every turn of the head, every item lying in the acoustic path, every cushion on the sofa, every wall hanging, every glass door in the wardrobe, every carpet, every room resonance and above all every single wall made of stone or a resonating wooden frame with damping filling makes a far more significant and, above all, audible and measurable difference than the difference between 90 and 120 db SINAD at an Digital Audio Converter.

Pushing the limits of what can be measured in audio technology out of the audible range makes sense, but only if the same effort is put into the fundamentals of room acoustics.

The room interacts with the loudspeaker and I would go so far as to say that the listening room is part of the loudspeaker itself.

That is true, that is why i don't mind changing my +10 years old cambridge Dacmagic (measured here) with a newer higher Sinad model. It does the job good enough and is certainly not the bottleneck for good sound. Even my even older Steinberg UR-22 interface won't be that (even if it's quiet noisy for today's dac standards). I use that steinberg now mainly as mobile dac, on places where it won't matter much

 

[ahofer]

The question of timbre and measurements came up, and was discussed at length elsewhere

What is timbre and can we measure it?

My understanding is timbre is describing the differences in sound of different instruments playing the same note. My understanding is also that this site effectively evaluates musical equipment by passing a set frequency (a note) and evaluating how faithfully the equipment reproduced that...

audiosciencereview.com

Like pitch, timbre’s definition in art has a lot to do with perception, above and beyond the signal itself. We can certainly measure the signal components, although i don’t know that we have the right standard ensemble of measurements to help the user correlate their perception of timbre accuracy to the signal input.

And, as raised earlier, there is the question of strict accuracy of timbre to the original sound, apart from any preference effects. Nonetheless, the *signal* that alters timbre perceptions lies within frequency and amplitude as reproduced over time. Right?

It seems to me that signal accuracy necessary to timbre accuracy within the signal chain from mic input to speaker terminal should be covered in the standard kinds of distortion measurements. Transducers have much larger inaccuracies within which a timbre-altering signal distortion might lurk.

Disagree?

 

[rdenney]

When it gets to the point where you can measure timbre directly, my guess is that it will transform speaker “measurements” and timbre will be incorporated in measurements in a meaningful way. You would be able to see, objectively and scientifically, that Speaker A reproduced a (fill in blank, instrument, jazz trio, female voice) by some measure of accuracy and that Speaker B reproduced that sound by some level of accuracy. If that measure of accuracy leads to a more precise determination of preference in speaker selection it will become part of the measurements.

It would need to be at that level of precision where it can ID a voice, or an instrument, to provide useful information (what coloration, and by what amount). In other words, it would have to take things beyond what we already have (frequency sweeps at various horizontal and vertical angles to predict a preference).

I'm troubled by hanging this entirely on "timbre". For me as a musician, timbre is the characteristic sound of a musical instrument, and it involves several factors: 1.) the spectral composition, 2.) the nature of the articulation, 3.) the nature of frequency or dynamic variation, and 4.) the propagation directivity in a space. I think these are in order of influence, but that's a guess. If we (as casual listeners) listen to a stack of sine waves that reproduce the spectral content of, say, a tuba, we still might not recognize it as a tuba. If we add the articulation--the way in which the tone is initiated in time and pitch--it becomes far more recognizable. If we add just the slightest bit of pitch and dynamic variation--not nearly to the extent of being recognizable as vibrato--it becomes human and distinguishes itself from a synthesizer. And if it fills a space in the right way, the sound mixes in a space in a way that localizes and identifies the source. The last is the reason why two instruments might seem distinct when heard live but not when recorded (at least not in mono or stereo), but I don't think it's at all necessary for recognizing the instrument for what it is.

But in terms of how those features interact with the room, I don't think that's as important to "timbre" though it may be important to "realism". I think anyone who has heard one before can identify a tuba no matter where it is played, from on the street with zero reverberation to in a church with an RTA60 of five seconds. Those room effects identify the room, not the instrument, because our brains know how to interpret room effects. Toole made this point clearly. If we have room effects much different than what could be possible in the listening room, it won't seem "live" but it might still seem "realistic," it seems to me. It also seems to me that if we want it to sound like the tuba player is playing the instrument in the chair next to the speaker, the recording needs the same acoustic space effects as the listening room.

I personally don't need that--I'm happy to let the recording transport me to the performance space rather than expecting it to transport the musician to my living room. This is the same willing suspension of disbelief that allows me to read well-crafted and evocative literature and feel as though I am experiencing what is being written about while sitting on my couch.

Rick "no way any speaker can propagate sound in a room like a tuba" Denney

 

[watchnerd]

So lets say for argument sake pavarotti and bon jovi hits an exact same note. It will measure the same wont it? Will it sound the same?

Not if you look at the waveform, it won’t look the same at all

 

[Ashl777]

This is all measurable and can also be made visible in various forms of visualization.

The problem is really not measurability.

All of the Field of Audio can be measured today, but you also have to be able to interpret it.

In the field of loudspeakers, the basic technology has basically not changed since the invention of electrodynamic loudspeakers by Edward Kellogg and Chester W. Rice in 1925.
A modern BMS or Purifi driver is based on the same principles of the spring mass system.
Evolutionarily improved, but the same principle - with all its flaws and measurable limitations.

There is some research in other directions, like MBL and Manger or the Plasma Ion Tweeter, but basically the speaker in most living rooms is the same old principle.

It would therefore always be necessary to measure the living room with the specific loudspeaker in order to arrive at truly enlightening measurements.
This is being practiced more and more with amplification and DSP support.

Incidentally, this has been common practice in PA technology for a long time.

In any case, I always have to smile when someone talks about a value of at least 120 Sinad and the situation then shows the great DAC in combination with two small loudspeakers in a resonating shelf.

Because every angle of change in speaker position, every change in distance, every turn of the head, every item lying in the acoustic path, every cushion on the sofa, every wall hanging, every glass door in the wardrobe, every carpet, every room resonance and above all every single wall made of stone or a resonating wooden frame with damping filling makes a far more significant and, above all, audible and measurable difference than the difference between 90 and 120 db SINAD at an Digital Audio Converter.

Pushing the limits of what can be measured in audio technology out of the audible range makes sense, but only if the same effort is put into the fundamentals of room acoustics.

The room interacts with the loudspeaker and I would go so far as to say that the listening room is part of the loudspeaker itself.

Very informative thank you. Maybe why some speakers in a certain environment doesnt sound as good as in other even if it measures better. I found room eq to be the biggest bringer of change and audio bliss in my 20years in hifi. Only realising it when inhad proper eq’d room with something like trinnov. Delving into the dark arts of room treatment is another……

 

[Holmz]

Very informative thank you. Maybe why some speakers in a certain environment doesnt sound as good as in other even if it measures better. I found room eq to be the biggest bringer of change and audio bliss in my 20years in hifi. Only realising it when inhad proper eq’d room with something like trinnov. Delving into the dark arts of room treatment is another……

Airplanes, compasses, radios and other tech always seems like the dark arts to the remote tribesman/tribesperson.

 

[Cbdb2]

Yes. Timbre-it can’t be measured directly (yet). Is Timbre real? Does it matter? Yes it’s real, it’s how you tell an oboe from a bassoon, a sax from a trumpet even if playing an identical note. It how a baby can recognize its mother’s voice from a room full of mothers.

Machines can also recognize the babys mothers voice, so timbre can be measured. FFT.

 

[watchnerd]

Machines can also recognize the babys mothers voice, so timbre can be measured. FFT.

My stock brokerage account has good enough voice authentication to be able to identify me by voice signature over the crappy phone mic.