What is timbre and can we measure it?

[Chr1]

Sorry, I don't think that this is true. Kind of a wild statement, and highly unlikely. Evidence please.

There's certainly no one Fender or Gibson sound. Which actual guitars?Audio mythology. WTF.

This thread continues to get weirder.

 

[Axo1989]

Whereas dictionary.com has:

And you'll note that as I've stated above I'm quite happy to accept the word applying to things other than instruments. I just think it is more commonly applied to muscal situations.

Do you feel my understanding of the term is too narrow? In what way?

That definition is more comprehensive, and includes variation according to two different professional domain contexts, which is an improvement. Wrt narrowness or otherwise of your understanding, I disagree with you at posts #594 and #630 respectively:

I think anything that produces sound can be said to have timbre IF its sound has characteristics that allow you to identify what the thing is that is producing the sound.

Identification isn't a sonic attribute. Differentiation maybe. I'm unlikely to identify different string or woodwind instruments, but I may be able to perceive different timbres thereof. I'm not interested enough in acoustic instrument (in and of themselves) to care about categorisation. And I may or may not realise that different sounds come from particular instruments. None of that precludes perceiving timbres from those sounds.

I think the key part of the definition is "...that allows us to distinguish between instruments"

That would be a consequence of how we perceive timbre, but no different for instruments vs other sounds (musical or otherwise).

To clarify, it was @solderdude who declared the instrument-only position, and their reply to me at post #601 that invoked an engineer's approach—as distinct from a scientist's—and led to my comment thereon. I agree you didn't entirely preclude other sounds, forgive any apparent conflation.

Getting back to dictionaries and definitions, my dictionary app offers considerable material on timbre, including the Erickson example I quoted upthread. Re-reading now I can see an ASA definition (per Acoustical Terminology ANSI S1.1–1994) also:

"that attribute of auditory sensation which enables a listener to judge that two nonidentical sounds, similarly presented and having the same loudness and pitch, are dissimilar", adding, "Timbre depends primarily upon the frequency spectrum, although it also depends upon the sound pressure and the temporal characteristics of the sound".

I glossed over that previously but see now the wording "auditory sensation" is different again (vs "instrument" or "sound") and consistent with @chervokas various posts re timbre as a perceptual phenomenon. Which is interesting.

 

[pablolie]

To go back to the original post, I think every possible definition of timbre has been provided. What nobody has stated ever is that timbre is a blind area in measurements, which was the real masked questions in the OP. I give you:

".. Secondly, is this picked up on our "normal set" of measurements?.. is this a potential area that measurements are lacking?.. "

All I am interested in this discussion is that everybody agrees that the answer to that is a universal, resounding "no"...

 

[Guermantes]

From Bregman's Auditory Scene Analysis (p.92 ->):

The problem with timbre is that it is the name for an ill-defined wastebasket category . . . I think the definition of timbre by the American Standards Association should be this: "We do not know how to define timbre, but it is not loudness and it is not pitch." . . . What we need is a better vocabulary concerning timbre. But the vocabulary should not be selected arbitrarily. The dimensions that are wanted should ideally have two properties: (1) they should act in psychologically simple ways (that is, they should act in independent or perhaps additive ways in controlling scene analysis), and (2) they should, if possible, have straightforward physical definitions. What we need to do is develop descriptive terms for timbre, find ways to measure them, and do careful studies of how they affect perceptual grouping. Until such time as the dimensions of timbre are clarified perhaps it is better to drop the term timbre and talk simply in terms of those features of sounds that can affect sequential integration.

 

[Chr1]

So I would suggest that what might be useful at this point is ignoring the fact that timbre is a perception and that only voices and musical instruments can impart it.

We are nerdy science types, so why not break down what timbre is, without the burden/complication of the perception bit. Or just call it something else, such that we can choose to define it. WTF. Doubtless this is wrong, but otherwise we're looking at an eternal thread of vagueness. Everyone's perception is different.

I would still like to know if all aspects of timbre/tone/new-word can be measured, and if so how. This was the OPs question. Howsabout forgetting the perception bit and just call it tone or something else.

For those in the know : What are the measurable attributes of timbre/new-word. I get that it is complex as time/duration is involved, but howsabout having a go?

What becomes timbre once perceived and how can we measure/quantify/define it?

 

[Axo1989]

So I would suggest that what might be useful at this point is ignoring the fact that timbre is a perception and that only voices and musical instruments can impart it.

We are nerdy science types, so why not break down what timbre is, without the burden/complication of the perception bit. Or just call it something else, such that we can choose to define it. WTF. Doubtless this may be seen as wrong, but otherwise we're looking at an eternal thread of vagueness.

I would still like to know if all aspects of timbre/tone can be measured, and if so how. This was the OPs question. Howsabout forgetting the perception bit and just call it tone or something? This would still be useful to me.

For those in the know : I'd still like to know what the measurable attributes of timbre are. I get that it is complex as time/duration is involved, but howsabout having a go?

So you are saying "there's this thing that I don't understand too well, can we discard most of it and measure as if it's a different thing I understand better"?

I can see the appeal there.

Jokes aside, I'd say we can measure all of the sonic aspects that feed into perceptions of timbre, but that it's a complex thing, so you won't get a simple number or index from that. Complexity and vagueness are different things, obviously, but they can be confused. Studying the perceptual aspects is a different discipline, but not intrinsically invalid because of that.

 

[Titurel]

I would say a fuzz box (or any audio processing device / effect) affects timbre, otherwise why use it? Saying it "has" a timbre would be overbroad per standard definitions.

But consider, a violin. Without physical input (e.g., arco, pizzicato, col legno) it has no timbre (or sound), just like the fuzz box. So, conceptually (transformationally), they are similar if not identical.

Play a trumpet with the bell an inch or so from the sound-hole of an acoustical guitar. The guitar rings. Both are creating their own sound, and the sound of each and of both of them together have timbre.

Why insist that sound itself cannot be used to induce (new) timbre from another coupled device or instrumentality? Does it matter conceptually if the coupling is acoustical or electrical? I don't believe it does.

 

[IPunchCholla]

So I would suggest that what might be useful at this point is ignoring the fact that timbre is a perception and that only voices and musical instruments can impart it.

We are nerdy science types, so why not break down what timbre is, without the burden/complication of the perception bit. Or just call it something else, such that we can choose to define it. WTF. Doubtless this may be seen as wrong, but otherwise we're looking at an eternal thread of vagueness.

I would still like to know if all aspects of timbre/tone/new-word can be measured, and if so how. This was the OPs question. Howsabout forgetting the perception bit and just call it tone or something else.

For those in the know : What are the measurable attributes of timbre/new-word. I get that it is complex as time/duration is involved, but howsabout having a go?
Ta. Muchas gracias

Why not just go with timbre is the frequency spectrum of a sound? That’s what’s left after you ignore perception. It is our brain processing the frequency spectrum of a sound that allows us to identify it as a violin or as THAT violin.

If one wants, one could break it down further and say it is the frequency spectrum of a sound paying particular regards to the 4 components of that sounds frequencies and its envelope. The frequencies have both patterned and unattended components which we can call harmonics and partials. The envelope has 4 stages (attack, decays, sustain, and release.

We measure timbre by plotting a sound’s frequencies frequency and intensity against time.

 

[Chr1]

So you are saying "there's this thing that I don't understand too well, can we discard most of it and measure as if it's a different thing I understand better"?

I can see the appeal there.

Jokes aside, I'd say we can measure all of the sonic aspects that feed into perceptions of timbre, but that it's a complex thing, so you won't get a simple number or index from that. Studying the perceptual aspects is a different discipline, but not intrinsically invalid because of that.

I agree but I am trying to establish what if anything we can measure and quantify from this rather vague term "timbre". If it is purely based on personal perception then perhaps this is pointless. I don't know. But the talk of Gibson vs Fender sound seems kinda ridiculous to me.

 

[pablolie]

So you are saying "there's this thing that I don't understand too well, can we discard most of it and measure as if it's a different thing I understand better"?

I can see the appeal there.

Jokes aside, I'd say we can measure all of the sonic aspects that feed into perceptions of timbre, but that it's a complex thing, so you won't get a simple number or index from that. Studying the perceptual aspects is a different discipline, but not intrinsically invalid because of that.

There is absolutely nothing about timbre that falls outside basic audio measurements and playback thereof. All the pages about its definition don't change that. Could also title the thread "does a stradivarius challenge our established audio physics?" and "can you define the sound of a stradivarius" and it'd be the same waste of time...

 

[kemmler3D]

But consider, a violin. Without physical input (e.g., arco, pizzicato, col legno) it has no timbre (or sound), just like the fuzz box. So, conceptually (transformationally), they are similar if not identical.

A violin needs physical input of energy to produce sound acoustically, while a fuzzbox actually needs an audio signal to produce sound. It can't produce sound on its own, usually. So conceptually similar but different enough not to be confusing, I think.

Let's just go straight to the worst (best) and most confusing example I can think of, the talkbox.

Where should we locate the timbre in a talkbox performance? (e.g.)

In my mind the timbre is attributable to the whole system, possibly including the vocal mic. I don't see any problem with that, but is there any point in also debating whether the compression driver, tube, or enclosure of the talkbox (or EQ / effects / mic) has its own timbre?

I agree but I am trying to establish what if anything we can measure and quantify from this rather vague term "timbre". If it is purely based on personal perception then perhaps this is pointless

Some audio phenomena have physical and perceptual counterparts. The physical one can be measured and used to predict perceptions, but not necessarily 100%.

Physical --> Perceptual
----------------------------
Volume --> Loudness
Frequency --> Pitch
Timing --> Rhythm
Harmonic content (plus other stuff) --> Timbre

We can measure everything that goes into timbre, but it's not a measurable quantity in and of itself, it's the word we use to describe our total impression of those various measurable quantities.

 

[Robin L]

I agree but I am trying to establish what if anything we can measure and quantify from this rather vague term "timbre". If it is purely based on personal perception then perhaps this is pointless. I don't know. But the talk of Gibson vs Fender sound seems kinda ridiculous to me.

Don't play guitar?

 

[Axo1989]

Why not just go with timbre is the frequency spectrum of a sound? That’s what’s left after you ignore perception. It is our brain processing the frequency spectrum of a sound that allows us to identify it as a violin or as THAT violin.

If one wants, one could break it down further and say it is the frequency spectrum of a sound paying particular regards to the 4 components of that sounds frequencies and its envelope. The frequencies have both patterned and unattended components which we can call harmonics and partials. The envelope has 4 stages (attack, decays, sustain, and release.

We measure timbre by plotting a sound’s frequencies frequency and intensity against time.

I'd agree. The sonic inputs to what we perceive as timbre are basically the frequency spectrum and time domain presentation of a given sound (or combination of sounds, as applicable). It's fairly straightforward conceptually.

I think conceptual complexity is imposed by arbitrary exclusions (instrument or not, music or not, and so on). Just don't do that.

Edit: and the aspect that @kemmler3D added to their post above: trying to force "timbre" to be a measurable quantity in and of itself.

 

[Chr1]

There is absolutely nothing about timbre that falls outside basic audio measurements and playback thereof. All the pages about its definition don't change that. Could also title the thread "does a stradivarius challenge our established audio physics?" and "can you define the sound of a stradivarius" and it'd be the same waste of time...

That said, I am still intrigued to know how it is measured though .
And surely this is crucial if trying to describe the sound of a particular instrument.
This is important surely.
Perhaps those in the know here could explain the measurements required.
Thanks.

 

[kemmler3D]

We measure timbre by plotting a sound’s frequencies frequency and intensity against time.

This is not different than saying a recording is a measurement of timbre, which I guess it is, but I don't know if this gets us anywhere.

Timbre is what makes instruments or voices sound like themselves, which of course is encapsulated in a complete measurement/recording of how they sound.

Some attributes of the sound are more important to timbre than others for certain instruments, but it's hard to generalize.

 

[Chr1]

So, if timbre is perceived and can't be measured, what aspects of the sound can be measured and used to define it...
... prior to going in ones ear and becoming timbre.

What makes the sound identifiable is the perception/memory/timbre bit perhaps

 

[MattHooper]

But consider, a violin. Without physical input (e.g., arco, pizzicato, col legno) it has no timbre (or sound), just like the fuzz box. So, conceptually (transformationally), they are similar if not identical.

Play a trumpet with the bell an inch or so from the sound-hole of an acoustical guitar. The guitar rings. Both are creating their own sound, and the sound of each and of both of them together have timbre.

Why insist that sound itself cannot be used to induce (new) timbre from another coupled device or instrumentality? Does it matter conceptually if the coupling is acoustical or electrical? I don't believe it does.

Along those lines…

Per Toole/Olive, where timbre derives from resonances: Take a guitar string stretch between two points and not attached to anything else and pluck it. It’s going to have a certain timbre.

Placed the guitar string on a resonating guitar soundboard. Now it’s going to cause the soundboard to resonate. The soundboard resonating has a timbre. The timber of the resonating string and the timber of the resonating soundboard are going to create a combined timbre.

Likewise, Toole talks about various resonances that can occur in loudspeakers, including cabinet residences, port resonance, driver resonance Etc..

Like the guitar string causes the wooden soundboard to resonate and we apprehend that resonance as timbre, likewise music played through a loudspeaker could excite the speaker resonances, which we might perceive as the timber of that loudspeaker.
One that is going to distinguish one loudspeaker playing the same pitch/loudness from another loudspeaker.


(The paper I cited earlier suggests that listeners did indeed perceive such differences between loudspeakers in tests)

 

[Axo1989]

So, if timbre is perceived and can't be measured, what aspects of the sound can be measured and used to define it...
... prior to going in ones ear and becoming timbre.

What makes the sound identifiable is the perception/memory/timbre bit perhaps

I suppose we could come up with something analogous to the biometric coordinates of an individual fingerprint? Then group them. Maybe.

 

[kemmler3D]

what aspects of the sound can be measured and used to define it...
... prior to going in ones ear and becoming timbre.

What makes the sound identifiable is the perception/memory bit perhaps

Based on what I remember from college, the big factors going into timbre are the gross harmonic structure of the instrument's sound (i.e. the loudest harmonics / partials), the overall envelope (ADSR), how quickly various partials decay after onset, and the mix of harmonic / inharmonic content in the sound.

You can probably find better references that go into detail on this.

These things are all measurable but not easy to quantify in simple numbers.

For example we can measure distortion as a series of harmonics that all have individual amplitudes. We can see each harmonic, but it's hard to say what this sounds like just by looking at it. To make things easier we sum this up into one number as THD.

A measurement of timbre is a lot like a measurement of THD in that you will come up with a bunch of components, each with different amplitudes that change over time. However, unlike with distortion, there is no straightforward and useful way to sum timbre up into a number or even a few numbers.

There will often be 50-100+ audibly important partials in a sound, all of which you might find are impactful for its timbre. And each can change amplitude every few milliseconds. And all of those individual tones varying moment to moment ARE the timbre. And you can measure all that just fine, but it's hard to generalize about it.

Timbre is just a word for the overall sound of an instrument/voice, so the simplest way to experience it is to listen.

If you would like to play with timbre a bit, I can recommend downloading FL Studio (free demo) and loading some sounds into the Harmor plugin. It breaks audio down into those components and lets you manipulate them quite a bit. You can also watch a real-time visualization of each partial coming out of it. So this will tend to illustrate the concept for you in real time.

 

[kemmler3D]

I suppose we could come up with something analogous to the biometric coordinates of an individual fingerprint? Then group them. Maybe.

You mean something like "Brass, woodwinds, percussion, keys, bowed strings, plucked strings..." ?