What is timbre and can we measure it?

[Titurel]

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.

You don't get it. The fiddle is simply kindling waiting for a match without some form of excitation. It's as inert and tuneless as a fuzz box. You're making the frightfully arbitrary decision by excluding any piece of instrumentation whose timbral source is not purely muscular. I guess we need to toss out synthesizers, too. Back when I owned a Hammond B2, I thought its timbre was appealing. Guess I was wrong. It couldn't have had timbre. Boy was I wrong when I saved up my pennies for a Leslie.

Think about brass instruments. Now blow a raspberry. If you were skilled you could play tunes with just your lipe. I've seen many brass player warming up with just a mouth piece, playing scales and tunes. By your lights, attaching a trumpet, tuba, or trombone to that mouth piece wouldn't be according to Hoyle. The instruments wouldn't produce timbre because they couldn't be activated except by sound,. That doesn't make sense, Sorry.

Your exclusionary factor would disquality many if not most instruments generally conceded to produce timbre.

 

[kemmler3D]

You don't get it. The fiddle is simply kindling waiting for a match without some form of excitation. It's as inert and tuneless as a fuzz box. You're making the frightfully arbitrary decision by excluding any piece of instrumentation whose timbral source is not purely muscular. I guess we need to toss out synthesizers, too. Back when I owned a Hammond B2, I thought its timbre was appealing. Guess I was wrong. It couldn't have had timbre. Boy was I wrong when I saved up my pennies for a Leslie.

Think about brass instruments. Now blow a raspberry. If you were skilled you could play tunes with just your lipe. I've seen many brass player warming up with just a mouth piece, playing scales and tunes. By your lights, attaching a trumpet, tuba, or trombone to that mouth piece wouldn't be according to Hoyle. The instruments wouldn't produce timbre because they couldn't be activated except by sound,. That doesn't make sense, Sorry.

Your exclusionary factor would disquality many if not most instruments generally conceded to produce timbre.

I get it, I just think the everyday categories of "instruments" and "effects" as commonly understood are good enough. Instruments are said to have "timbre" and effects have a "sound" and that seems to work okay. I agree there is no obvious rational basis for the distinction at a physical level, I know it's arbitrary, but I also don't have a problem with that.

Just like shampoo isn't normally said to have a "flavor" even though it really does if you put it in your mouth, non-instruments don't need to have a "timbre" in my book.

 

[Axo1989]

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

Haha, a bit redundant, then? Less redundant when considering sonic range beyond traditional/unmodified acoustic instruments though.

I was thinking of analogous methods where metrics were extracted from graphical information (like a fingerprint) that might be applied to a wavelet spectrogram (for example). Fingerprint biometrics aren't notably human-readable, but something along those lines could potentially satisfy people who want to see abstracted number/index (I'm not really one of those people).

I get it, I just think the everyday categories of "instruments" and "effects" as commonly understood are good enough. Instruments are said to have "timbre" and effects have a "sound" and that seems to work okay. I agree there is no obvious rational basis for the distinction at a physical level, I know it's arbitrary, but I also don't have a problem with that.

Just like shampoo isn't normally said to have a "flavor" even though it really does if you put it in your mouth, non-instruments don't need to have a "timbre" in my book.

Oh I disagree with you there, of course. In the sense that any range from instruments to effects is a continuum with no obvious inflection point (also instruments have a sound and effects have timbre is just as true). So why distinguish when thinking of timbre ...

 

[pablolie]

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.

It's measured by our ears ability to hear between 20Hz to 20kHz. And our equipments' ability to play it back accurately.

Newton's law doesn't change whether it's an apple or a pear or a coconut or xyz hitting you on the head. (You may ask for a definition of xyz, but if it has a mass and is exposed to earth's gravity it don't matter :-D).

I retire from this discussion.

 

[Axo1989]

It's measured by our ears ability to hear between 20Hz to 20kHz. And our equipments' ability to play it back accurately.

Not really a measurement, more of a range.

Newton's law doesn't change whether it's an apple or a pear or a coconut hitting you on the head.

If it's my head, I'm voting for (over-ripe) pear.

I retire from this discussion.

There's no shame in that.

 

[pablolie]

Not really a measurement, more of a range.

A range we can accurately measure, and which is the only range that matters as far as our ability to hear music, timbre or a mouse farting behind the wall.

 

[Guermantes]

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.

Yes, Bregman points out that this "plus other stuff" is difficult to reduce to a fixed set of parameters, so timbre might best be considered a conceptual container (or "wastepaper basket") for a varying collection of acoustic attributes. For example, some sounds might be considered to be made up of "acoustic grains" and could be described effectively with granular synthesis parameters. Some sounds may be metameric in that they have different spectral distributions and yet are perceptually equivalent, but what is the "other stuff" that might produce that similarity.

 

[Titurel]

I get it, I just think the everyday categories of "instruments" and "effects" as commonly understood are good enough. Instruments are said to have "timbre" and effects have a "sound" and that seems to work okay. I agree there is no obvious rational basis for the distinction at a physical level, I know it's arbitrary, but I also don't have a problem with that.

Just like shampoo isn't normally said to have a "flavor" even though it really does if you put it in your mouth, non-instruments don't need to have a "timbre" in my book.

The problem is, you can't really define what is or isn't an instrument. Neither can I, and I played half dozen or so over the course of 50 years in the music business.

Is the Golden Gate Bridge an instrument? I heard it used as one. (samples). I had a couple of friends who played the saw. (There was some heated argument about whether rip or cross cut was the preferred form.) Hambone uses the body as an instrument. Tap shoes are musical instruments A friend of mine programmed a computer back in the 60s to play tunes by controlling the speed of the tape drives. Are the cannons in the 1812 overture instruments? Their timbral output is indicated in the score. Ring modulators (and the like) open a large can of worms in terms of what is or isn't an instrument and thus what is or isn't capable of creating timbre(s).

Every timbral source transforms one sort of energy or another into sound, whether the energy is black powder or the output of a Fender Stratocaster. It's really not any more difficult than that.

Addendum: Following your taxonomical rule to its logical conclusion, Jazz is not music since it most frequently relies on other music for its genesis. It's sound (and fury), not music.

 

[Axo1989]

A range we can accurately measure, and which is the only range that matters as far as our ability to hear music, timbre or a mouse farting behind the wall.

Adds nothing to the subject though. We can hear timbre, by definition.

But the actually sad part is I categorically ignore people who make (gratuitous) fart references. Oh well, it's been real ...

 

[Chr1]

It's measured by our ears ability to hear between 20Hz to 20kHz. And our equipments' ability to play it back accurately.

Newton's law doesn't change whether it's an apple or a pear or a coconut or xyz hitting you on the head. (You may ask for a definition of xyz, but if it has a mass and is exposed to earth's gravity it don't matter :-D).

I retire from this discussion.

Trouble is, our ears are probably the last thing (literally) you should use to measure audio.

No disrespect, but not sure what Newton has to do with audio either.
However, I vote for the apple having timbre.

 

[Axo1989]

Trouble is, our ears are probably the last thing (literally) you should use to measure audio.

No disrespect, but not sure what Newton has to do with audio, but I vote for the apple having timbre.

Well @MattHooper wouldn't mix up his apple/pear/coconut sonics when assembling that soundtrack.

 

[Chr1]

Perhaps timbre is like exotic cocktails. Difficult to describe, define and remember (with measurements) yet clearly subjectively enjoyable.
Then later, still really enjoyable. Tho less clearly

 

[kemmler3D]

In the sense that any range from instruments to effects is a continuum with no obvious inflection point (also instruments have a sound and effects have timbre is just as true). So why distinguish when thinking of timbre ...

In software it's easy to draw a line - one makes a sound, one doesn't do anything without an audio signal.

Isn't that a good enough distinction in real life too? Some things produce a sound of their own when you use them and others don't.

you can't really define what is or isn't an instrument.

I'll try anyway - an instrument is anything someone uses to produce sound for music. Conversely, anything that can't be used to originate a sound can't be an instrument.

Golden gate bridge - instrument, if you want it to be.

Fuzz box - not an instrument because you couldn't produce sound with just the fuzz box.

Every timbral source transforms one sort of energy or another into sound, whether the energy is black powder or the output of a Fender Stratocaster. It's really not any more difficult than that.

Agree. Effects don't convert energy into sound, they just do something to a sound.

Ring modulators (and the like) open a large can of worms in terms of what is or isn't an instrument and thus what is or isn't capable of creating timbre(s).

Modulators (as in, just the modulator) aren't instruments because they need something to modulate. An FM or RM synth of course is an instrument. This is why I think the talkbox is a tricky example because the vocal tract acts as a filter / modulator in contrast with its normal function as the instrument.

 

[sonitus mirus]

Perhaps timbre is like exotic cocktails. Difficult to describe, define and remember (with measurements) yet clearly subjectively enjoyable.
Then later, still really enjoyable. Tho less clearly

If I had to make an analogy, the way you reference timbre is similar to how one might refer to taste. While we can measure the ingredients and chemical composition of a dish, whether it tastes sweet, sour, spicy, bitter, gross, or delicious is subjective. What is taste, and can it be measured? Can the material and shape of a fork or chopsticks alter the taste? It seems like the same rabbit hole.

 

[Chr1]

Yes, but with good cocktails.

 

[Axo1989]

In software it's easy to draw a line - one makes a sound, one doesn't do anything without an audio signal.

Isn't that a good enough distinction in real life too? Some things produce a sound of their own when you use them and others don't. ...

I follow the logic for sure but it's a bit messy?

Say an analog synth has oscillators (an instrument) but also filters, modulators etc (not an instrument). So it's an instrument, or not entirely an instrument, or a compound instrument/effects rig, etc. Or a violin has strings (an instrument) but also a resonator (not an instrument). And so on.

If we simply say the timbre is an attribute of auditory sensation, that we derive from a (usually complex) sound—and that sounds have certain spectral and time-domain characteristics which can approximately predict perceived timbre—then we don't have to navigate all those qualifying hoops and cut-de-sacs.

Edit: just noting that only the first quote in that post is me, the others are @Titurel I think

 

[Chr1]

Yes. I reckon that electronic instruments such as synthesisers are going to be the hardest to define due to their complexity. Acoustic instruments perhaps.
Bell, Cowbell, Too much Cowbell, etc.

 

[IPunchCholla]

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.

I'm not sure that it is, exactly. What I am talking about is this:

Is that a measurement of 1 second of audio captured by DI from an electric bass? Or is it a recording of it? both? Doesn't it describe the attack, decay, sustain, release, and frequency spectrum of a sound? While it is not a single number, it can be, and in facts is, a numeric representation of what an electric bass (with Big Muffin) sounds like. You can train neural networks to recognize the similarities of the patterns of different electric basses (the distinctive relationships that are present in the spectrum and the patterns those relationship make as they decay) in order to "understand" what is electric bass with Big Muffin, and what is not. To the point that there any number of tools you can use to extract or isolate electric bass with big muffin from a recording of a number of instruments playing simultaneously.

 

[Guermantes]

Pierre Schaeffer also had a problem with "timbre" (note the similarity to Bregman's "wastepaper basket" in the metaphor of the test tube residue):

Now, the concept of timbre seems precisely to be the least capable of being abstracted,
designating as it does the qualitative residue that remains at the bottom of the psycho-
acousticians’ test-tube once they have broken sound down into three [52] measurable
parameters: frequency, amplitude and duration. A qualitative residue which, as we have seen,
cannot be reduced to the straightforward perception of a characteristic spectrum of
frequencies.
The T.O.M., therefore, prefers the hypothesis of calibrations of criteria to the
experiment of Klangfarbenmelodie. One criterion, such as grain, could perhaps be abstracted
from the sound which bears it, whereas timbre represents only the overall perception of a
structure of criteria, which defines the personality of a sound object or instrument in relation
to others. (see CALIBRATION, 18).
If we want to leave the traditional system and devote ourselves to researching an
experimental system, we must give up the concept of timbre, which is too vaguely defined,
and stop re-enlisting it as a value, “out of nostalgia for pure music”, as the supporters of
Klangfarbenmelodie have tried to do.
So the concept of timbre, too coloured by its traditional meaning, is replaced by the
more general concept of characteristic or genre, and the more subtle concept of criterion.

From Michel Chion's Guide to [Schaeffer's] Sound Objects.

https://monoskop.org/images/0/01/Chion_Michel_Guide_To_Sound_Objects_Pierre_Schaeffer_and_Musical_Research.pdf

 

[kemmler3D]

You can train neural networks to recognize the similarities of the patterns of different electric basses (the distinctive relationships that are present in the spectrum and the patterns those relationship make as they decay) in order to "understand" what is electric bass with Big Muffin, and what is not. To the point that there any number of tools you can use to extract or isolate electric bass with big muffin from a recording of a number of instruments playing simultaneously.

I guess I would say the neural network's assessment is the "measurement" of timbre, but the spectrogram is just another way of representing the whole recording... which isn't a model or measurement of the timbre, it's the whole shebang.

I do sometimes say recordings are "measurements", but I think "a measurement" of timbre would quantify it somehow, here we've just visualized the whole sample.