Why EQ a sound doesn't change timbre?

[Gamma-UT]

I believe timbre is a somewhat subjective term.

Lets say you have a piece of wood and piece of metal.
When you hit them you hear what is what. If you EQ them you will still hear what is what, although the frequencies are altered. Depending on the amount of EQ I would call it a change on timbre aswell, but the "origin" of the sound and so the differences between wood and metal would still be distinguishable (to a certain point perhaps).

2 cent.

Agreed. Perception is critical to timbre. My hypothesis is that if you play a random listener an entirely synthetic sound and alter the EQ they will likely perceive a change in timbre much more quickly than if it a sound from a recognisable object, such as a bell or guitar string. The bell I would guess changes perceived timbre with EQ much more quickly than with the guitar or cymbal, because the brain is spending a lot of its effort on pitch and the relative harmonics. The brain might not notice inharmonic overtones in the bell until some of the louder, lower tones are removed.

Someone was doing a research project and posted some tests here a month or two back that I think was related to perception of timbre. I'm sure there's literature on timbral perception but haven't looked recently.

[Nowhk]

Later on we have established that EQ can change the timbre.

Yes, although someone doesn't agree with this:

No, timbre is perfectly "preserved" in any context of interest...
...microphone, an EQ, an amplifier, a room...

But go ahead

Start with putting all the knobs at neutral position, so sound passes straight through it and it is not filtered. The timbre of the sound thus stays the same. Correct?

Sure Until you change speakers... but lets go ahead

Other experiments have already shown that the human hearing cannot reliably discriminate volume differences smaller than 1dB. So if I take the 20kHz slider (highest frequency) and put it down by 0.5 dB, I am fairly sure that you and I cannot really notice the difference. So there's a tiny bit of EQ applied, but timbre did not change.

I can drop the 20kHz slider again by 0.5 dB, and we all should not notice the difference with the previous situation.

Yeah, the same here

But my son with young ears will notice

Stay with you, not referring to other people
Yes, I still don't notice differences without the 20khz band.

We can continue with the slider before (16kHz) and do the same thing. And the next band (12.5kHz) and the next (10kHz) and the next (8kHz). Now it sounds like the audio is travelling through a telephone line. You did admit previously that now the timbre has changed.

That's extreme ratio. Try put all the bands to neutral position again and just edits some bands in the middle up/down to some db, which is what happening on a speaker frequency response, more or less. Here's how you should place band slider emulating a NS10 (of course, its just an example, the differences are huge). Just a practical example!
Would you say the timbre has changed now?

So is it the question: where do we draw the line? At what point is it the same, or is it slightly simular, or is it different? With each tiny step it sounds very simular to the previous step. Yet when adding all the tiny steps added together, somewhere there is a point where both you and I will agree the timbre is different.

Yeah, there's a line between me and you, of course! But we both agree that different setup "CAN" change the timbre. I don't get the differences between "it sounds very simular to the previous step" and "is different". If it's not the same (but similar), it's different however. Once it's "similar", than its not "equal" to the original, thus its different.

My point proven: whether timbre changes depends on how much EQ and which listener you ask.

Perfect. I agree at 100%. So the "preservation" of timbre is still subjective. But may change!

Differences are there, but that should not be relevant.

This is somethings I can't get, really
Maybe my brain is so surgical, or maybe because I deal lots with synthesized sound... I don't know! But hearing a "bass" with two speakers able to play the same range (the human perceptible one) but with different shape (fr) and "ringing", give it to me two different results. The differences are "relevant". That's much more noticeable on kick and their transient. Club speaker or loudspeakers, vs monitor or headphones. At the end, I enjoy both, but I got them different.

You deal with it: find a compromise that works well in any situation.

I wrote somethings similar some post ago! Again: I agree at 100%! But still you can't say that timbre is 100% preservable (as it is for pitch or rhythm for example): compromise imply to referencing on somethings "similar" (or different, which how I would call them).
At the end you enjoy the sound "as it sound" for the fixed setup, there is not an absolutism.

If you EQ them you will still hear what is what, although the frequencies are altered.

Maybe you are a fresh reader, but again: I'm not talking about confuse a "guitar" with a "piano" after EQing or playback it on different setup (that's of course won't happens, prior to extreme scenarios), but to listen a "different" (similar) metal/wood.

Resuming (as I said some posts before): the target is (how BertKoor said, I guess) to find a good compromise (from sound design to mix/master) of making your track, in a way that, even if it won't play "equally" on your target's setups, it will be differently enjoyable (or similarly enjoyable, as you prefer ).

Heresy?

[slipstick]

Heresy?

Oh dear. No that's not heresy. It's called mastering and doing it well so that the sound translates onto as many possible systems as practical is why good mastering engineers are paid a lot of money.

Steve

[BertKoor]

Try put all the bands to neutral position again and just edits some bands in the middle up/down to some db, which is what happening on a speaker frequency response, more or less. Here's how you should place band slider emulating a NS10 (of course, its just an example, the differences are huge). Just a practical example!
Would you say the timbre has changed now?

That depends...

If you are listening to NS10 and another monitor side by side comparing: yes you clearly hear the difference in timbre between these speakers.

But if I hear a song on NS10 speakers one day and the next day the same song on any other random speakers: no it's probably the same timbre of the song.

That's your brain messing up. You need some reference to hear such differences, otherwise you will compensate.

And that is why it is so important that when shopping for monitors you go out to a real shop and listen to them. Not only for a pleasing sound, but also for revealing everything and not hyping bass / treble too much.

[ghettosynth]

I believe timbre is a somewhat subjective term.

Lets say you have a piece of wood and piece of metal.
When you hit them you hear what is what. If you EQ them you will still hear what is what, although the frequencies are altered. Depending on the amount of EQ I would call it a change on timbre aswell, but the "origin" of the sound and so the differences between wood and metal would still be distinguishable (to a certain point perhaps).

2 cent.

Agreed. Perception is critical to timbre. My hypothesis is that if you play a random listener an entirely synthetic sound and alter the EQ they will likely perceive a change in timbre much more quickly than if it a sound from a recognisable object, such as a bell or guitar string. The bell I would guess changes perceived timbre with EQ much more quickly than with the guitar or cymbal, because the brain is spending a lot of its effort on pitch and the relative harmonics. The brain might not notice inharmonic overtones in the bell until some of the louder, lower tones are removed.

Someone was doing a research project and posted some tests here a month or two back that I think was related to perception of timbre. I'm sure there's literature on timbral perception but haven't looked recently.

This requires that one get pedantic about definitions. Timbre is often defined as the "perceived" tonal quality of a sound, hence, talking about the perception of timbre is a bit redundant. That said, we commonly use the word to mean the spectrum over time, perceived or not, intrinsic in a sound. We might say "the timbre of the clarinet is rich in odd harmonics." In that sense, the timbre of a sound is independent of perception and can be thought of as spectrum measured by some ideal instrument with infinite bandwidth/input impedance.

So, when I say that the timbre of a recording is preserved modulo the transfer function of the systems that it passes through, I mean it in the same sense as if we say "the timbre of a clarinet is rich in odd harmonics", we don't actually have to listen to the clarinet. This is a statement of the properties of the timbre independent of human perception. The timbre of any played sound however, preserved as described above, arrives at your ears and then passes through the HRTF (head related transfer function) and then what might be called your BRTF (brain related transfer function) which is largely unknowable at any point in time for any particular person. While we can obtain a good approximation of the HRTF, the best that we can do for the BRTF is to obtain a crude statistical approximation that is very unlikely to acknowledge all sources of bias or anomalies such as momentary auditory hallucinations.

[Gamma-UT]

Timbre is often defined as the "perceived" tonal quality of a sound, hence, talking about the perception of timbre is a bit redundant.

That is true. Better to say "if we assume timbre is a perceptual quality and not a function of measured frequencies".

However, in the latter case, it's trivial: any EQ change other than gain is going to change timbre.

The timbre of any played sound however, preserved as described above, arrives at your ears and then passes through the HRTF (head related transfer function) and then what might be called your BRTF (brain related transfer function) which is largely unknowable at any point in time for any particular person. While we can obtain a good approximation of the HRTF, the best that we can do for the BRTF is to obtain a crude statistical approximation that is very unlikely to acknowledge all sources of bias or anomalies such as momentary auditory hallucinations.

This is the core of the issue: the brain adapts to try to make sense of what it encounters and so will work hard to fool the senses – like taking a photograph on a sunny day. You don't notice how dark the shadows are until you look at the 2D shot because the brain is continually compensating for the effect. A trained photographer compensates for it explicitly using fill flash but usually only after bitter experience of failure many times before.

With hearing, if you have a cold, your hearing will be different. But your brain is darn sure it's still a clarinet (though it might suddenly think, "that's an alto not a soprano", if the brain has learned the difference). That's why I think the issue of changes in timbre will depend heavily on what the brain has learned about sounds it thinks are made by real objects vs sounds made by unreal objects.

The experiment would depend on listening tests across a large group, possibly divided into musicians/mixers vs general public.

[BertKoor]

I think I found out what might cause the confusion: keep it simple! What works out all the time when figuring out problems, is change just one thing and observe the difference. All other things must remain stable. This is because we are better at determining relative levels (differences) than absolute levels (measure without any reference)

So either:

1) you keep monitors etc the same but switch the played sound. You now can only observe timbre changes between the played sounds. It might happen that a visiting listener critiques the setup of the host, but after some time you get used to imperfections (or get irritated and leave)

2) you use the same sound to compare monitors side by side. You now only observe differences between the monitors timbre.

And this is why differences between speakers are not relevant if you talk about the timbre of the played sound. It is often not feasible to switch speakers, so these are taken out of the equation.

To make a parallel with visuals: white paper is always white. On a sunny day, on clouded day, even in the dark.

[Nowhk]

Lots of good material yesterday to thinking about... thanks guys for still partecipating

If you are listening to NS10 and another monitor side by side comparing: yes you clearly hear the difference in timbre between these speakers.

But if I hear a song on NS10 speakers one day and the next day the same song on any other random speakers: no it's probably the same timbre of the song.

I have to admin that this is really true. Didn't see by this point of view...

This requires that one get pedantic about definitions. Timbre is often defined as the "perceived" tonal quality of a sound, hence, talking about the perception of timbre is a bit redundant.

It could be that I totally confuse "timbre" (human perception thing) with "spectrum" (physical stuff). So even if the spectrum change, brain don't care, and with its own memory and experience, compensante all of the infos it get creating something "concrete" for me. Do you mean this with "preservation" of timbre?

So, when I say that the timbre of a recording is preserved modulo the transfer function of the systems that it passes through

What do you mean with "preserved modulo the transfer function of the systems"? I'm not able to translate this phrase.

I think I found out what might cause the confusion: keep it simple!
...
2) you use the same sound to compare monitors side by side. You now only observe differences between the monitors timbre.

And this is why differences between speakers are not relevant if you talk about the timbre of the played sound. It is often not feasible to switch speakers, so these are taken out of the equation.

To make a parallel with visuals: white paper is always white. On a sunny day, on clouded day, even in the dark.

Damn, thats true.

I'm totally re-evalutating my whole idea about music and "actors"...

- artist do "what" it want to express: elements such as harmony, melody, rhythm, timbre, ecc;
- producer/sound engineering do "how" they will be expressed: mixing/mastering and so on;
- listener choose its ideal setup that will play the whole work in his prefered way, getting "what" in the better way due to "how" it is packaged;

If that's true, this create an "abyss" on my whole understanding of music. I've always defined a mixdown part of the song; surely my brain was working in the right way (extrapolating e consuming music as it is), but my reasoning was totally wrong.

In this way, reverb/eq/compression/ecc on mixing/matering stage (of course used in a moderate way, not shaking up the whole work) are not really part of the "music" itself, they are just tools for packaging and expose in a "good way" the "what" that artists would like to express.

What do you think about this reasoning? Paradoxically my first "claim" of this topic was true: EQ doesn't change timbre (as perceptive element).

Another funny thing (for me): if you were asking to me 2 days ago what do you think about this:

https://www.youtube.com/watch?v=gcbwH00IElU
I would say "I'm listening 3 different song", with 3 different timbres of the guitar.
Now actually I would say that the timbre of that guitar... well... its the same

[Nowhk]

So do you agree that on Tue May 30, 2017 22:23 you have admitted you understand EQ can change the timbre. Right?

That is true. Better to say "if we assume timbre is a perceptual quality and not a function of measured frequencies".

However, in the latter case, it's trivial: any EQ change other than gain is going to change timbre.
...
With hearing, if you have a cold, your hearing will be different. But your brain is darn sure it's still a clarinet (though it might suddenly think, "that's an alto not a soprano", if the brain has learned the difference). That's why I think the issue of changes in timbre will depend heavily on what the brain has learned about sounds it thinks are made by real objects vs sounds made by unreal objects.

I think that at this point is correct to say that "EQ change timbre" (as well as environment/setup/playback/and so on) only if we consider it as "physical" thing (i.e. the spectrum).

Also, EQ change the "hearing's timbre" (sensation), not the listening's one (perception, derived from sensation, memory, experience, bias and other factors).

Does this make sense for you? I'm reading Tuning, Timbre, Spectrum, Scale by William A. Sethares...

[Nowhk]

Hi there.

I'd like to post another (extreme) example, because I'm still reasoning on this.
From The Matrix Blu Ray VS DVD Side by Side Comparison 1080p:
Here's an extrapolated image at that frame:


Watching these two images, (for me) it seems that Neo got a different "mood" on that scene: I would say "worried" to the right side and something "astonished" to the left.

My perception DO "change" from VHS to DVD. In a light way, I know, but it does.
Here we are talking about different mastering, obviously, but I guess the same can be applied (in small doses) also watching on different environments/setups.

The "mood" that Keanu Reeves and Wachowskis would trasmit for that scene (for a fixed person like me) are "also" dictated by the "vehicle" your are using to look at it. And I'm not talking only about mixing/mastering, but also using different "same-level-quality" setups.
They have not a total (what I called concreteness/preservance) control over it. Some parts (and thus perception) it seems to vary naturally...

EDIT: Lol, after seeing the YouTube preview above, I would says that Neo's "mood" there to the left side is sleepy!

[BertKoor]

I see no difference in emotion, pure a technical difference. The picture on the left is too red, while the one on the right is too green.

I'd say that whoever did the color correction on either simply goofed up a bit.

[Nowhk]

I see no difference in emotion

Wow This is even more noticeable:


Different "color" balance and I got two different Morpheus's expressions.
The tunic looks more fashion and expensive on the right: quality cloth easy to dry after washing it.

I really perceive different characters...

[Nowhk]

However it appears you are still making two fundamental assumptions about the 'perception of it'

Firstly, you are assuming that the 'environmental changes' are significant enough to be perceived in the first place.

Secondly, you are assuming that your brain responds significantly to any such perceived 'environmental changes' so as to respond them differently, instead of, say, just filtering them out.

I'd be curious as to what your reference for perceiving the changes is, though. Some expectation that you have a perfect memory of the original, and that your brain is capable of some sort of discriminatory process of comparing one sensory stream against a memory of an incredibly similar sensory stream?

Do you think your brain maintains a perfect memory of every sensory experience you've ever had?

Do you know any ABX test I can try myself which proof that "environments changes" of a playback song won't affect the way I perceive its elements (such as heavily discussed timbre)? So I can confirm its just a bias of my mind...

I've read a lot about this concept these days. Its called "perceptual consistency" for what I get, but hey: its seems purely a theoric concept, nothing more.

Since you seem to trust in this as approved scientific thing, I would like to get rid of my uncertainty in first person. Fancy test? Books/articles?

[jancivil]

Timbre is often defined as the "perceived" tonal quality of a sound, hence, talking about the perception of timbre is a bit redundant.

That is true. Better to say "if we assume timbre is a perceptual quality and not a function of measured frequencies".

However, in the latter case, it's trivial: any EQ change other than gain is going to change timbre.

The timbre of any played sound however, preserved as described above, arrives at your ears and then passes through the HRTF (head related transfer function) and then what might be called your BRTF (brain related transfer function) which is largely unknowable at any point in time for any particular person. While we can obtain a good approximation of the HRTF, the best that we can do for the BRTF is to obtain a crude statistical approximation that is very unlikely to acknowledge all sources of bias or anomalies such as momentary auditory hallucinations.

This is the core of the issue: the brain adapts to try to make sense of what it encounters and so will work hard to fool the senses – like taking a photograph on a sunny day. You don't notice how dark the shadows are until you look at the 2D shot because the brain is continually compensating for the effect. A trained photographer compensates for it explicitly using fill flash but usually only after bitter experience of failure many times before.

With hearing, if you have a cold, your hearing will be different. But your brain is darn sure it's still a clarinet (though it might suddenly think, "that's an alto not a soprano", if the brain has learned the difference). That's why I think the issue of changes in timbre will depend heavily on what the brain has learned about sounds it thinks are made by real objects vs sounds made by unreal objects.

The experiment would depend on listening tests across a large group, possibly divided into musicians/mixers vs general public.

I don't agree really at all that the term is this subject to consensus. I'm aware that a definition of it is 'the perception of tone color' or what-have-you. So, you can say that with a cold an individual may not hear as well but a clarinet is still coming across as a clarinet. But what about the next individual who thinks a soprano sax is a clarinet? There is someone who surely will even if they know there is a difference, because their perception isn't fine enough. So now that bet is off. Is there not a measurable difference that now defines the timbre of one versus the other? What is the useful definition in the end?

[jancivil]

I think that at this point is correct to say that "EQ change timbre" (as well as environment/setup/playback/and so on) only if we consider it as "physical" thing (i.e. the spectrum).

The wiki on 'timbre' does this:
timbre (/ˈtæmbər/ TAM-bər, also known as tone color or tone quality from psychoacoustics), is the perceived sound quality of a musical note, sound, or tone.

then:

The physical characteristics of sound that determine the perception of timbre include spectrum and envelope.

If it is determined by measurable phenomena, it cannot be subject to much varied interpretation or we don't really have a definition, as far as I'm concerned. I don't like that. So, out of the word spectrum there I made a link to a definition of spectrum that's more specific.