Can you hear formants? Should you?

[TimR]

This is prompted by an argument over hearing overtones on wind instruments, but I thought I'd ask hear (here) <humor> about singer's overtones.

As I understand it, a formant is an overtone for a singer that can be modified by altering the vowel until it is properly reinforcing the pitch.

So it seems to me that what you listen for is the vowel sound, rather than hearing the formant as a distinct pitch.

I do hear a distinct overtone with throat singing, but not with normal voice.

So, can you hear the actual pitch of a formant? If so, is it worth learning to isolate them and hear them? Or does everybody listen for vowel, brightness, projection, etc., but not the overtones themselves?

Not sure I'm explaining this well, sorry.

[Felipe Carvalho]

Tim,

Formants are present on your voice during speech, not just singing. And the vowel is exactly the result of our perception of them (mostly F1 and F2).

So if you are hearing a vowel, you are hearing it. If you hear a low vowel, such as AH, it means the F1 freq is high. If you hear a high vowel, such as OO, it means that F1 freq is low. If you hear a front vowel like EE, it means that F2 freq is high. If you hear a back vowel like OO, it means F2 freq is low.

[Owen Korzec]

Yes you should hear formants but less as distinct pitches and more as tonal qualities

Just as you should hear vowels but not as speech vowels but shades of vowel color

Whenever you get as exact as language vowels or tuning to harmonic pitches you've probably missed the mark. Most correct vowel modifications will sound like they're a weird mix of vowel shades and a specific formant/tonal color that results in a tone that simply sounds and feels right - usually where one overtone is dominant but the others are still supprting it strongly too and balancing out the tone. It really is a balance though because its not always desirable to line up formants with harmonics perfectly. That's mostly if you're like holding out a note or accentuating a high note and want it stable and badass sounding. For fast moving melodies and lyrics you may often have to compromise formants for stuff like diction, continuity of vocal tract shape, naturalness of sound, etc

Does that make sense?

[TimR]

I hear vowels, I don't hear F1 and F2 as distinct pitches.

Similarly, I hear good or bad tone from a trumpet or clarinet player, I do not hear the actual overtones that make up tone as distinct pitches.

It has been suggested that I should be listening for distinct overtone pitches, and I am not sure that makes sense for either voice or wind instruments. I wanted to confirm that vocalists do not listen to distinct formant pitches.

But I still wonder - can you, if you try?

[Xamedhi]

I don't know if I understand what you are asking, but I think I do.

Listen to this short phrase and tell me if you hear something weird going on in the last word. I say "Quiero estar" in spanish.

https://app.box.com/s/cfvjvh1rysvtdiy7h8cv

What I hear, is an underlying pitch, very sharp and metallic. Probably an A#, but I don't know in what octave.

Can you hear it ? It is easier to hear it after you play the clip some times in a row.

[TimR]

Like a low rumble, underneath the pitch you are singing?

Maybe. It might be there, I think I hear it. You could load it into Audacity and look at the spectrum.

Interesting.

I was thinking all formants are above the pitch, maybe not.

[Xamedhi]

I went into http://www.virtualpiano.net/ , lol and the pitch I hear I think is the last A# to the right in the keyboard.

I am listening with headphones, so probably it is easier to hear.

[TimR]

Oh. Probably an artifact of my system here, I'll try again on different speakers.

[Owen Korzec]

Generally speaking there's no such thing as formants below the pitch - well in rare cases that happens (a very high pure ee for instance) but it wont be heard anymore since there is no such thing as harmonics below a pitch.

Yes you can hear formants as distinct pitches of harmonics amplified but you don't really have to to learn good tone

[Felipe Carvalho]

I hear vowels, I don't hear F1 and F2 as distinct pitches.

Ok.

A "formant on itself" (whatever that means ) sounds pretty much like a Wah-Wah effect on the guitar, lower freq sounds more "closed", higher, more "open". (You will perceive it as a vowel articulation, usually from oh to ah).

You will have trouble hearing it as distinct pitches because although the resonance of the vocal tract can be adjusted freely, the source harmonic content is the fundamental + multiples of the fundamental, so the formants on the sound spectrum will always be a function of the fundamental. And our brain process this as a single sound (human voice).

Its possible to hear details, but basicly what you are hearing is if the vowel is open or closed, or, if its forward or back. Thus different vowels.

About throat singing thats a different thing entirely. I am frankly not very familiar with it, but I suspect a big part of the deal is having a very non-linear source (phonation with some kind of distortion) to provide harmonic content that is not a function of the fundamental (all the samples I ever heard are distorted in some way).

[Owen Korzec]

There is a way of throat singing (the clean kind) that is literally just formant tuning but you have to double-amplify the formant by aligning F2 AND F3 with the harmonic. At that point the overtone is strong enough to be perceived as a second note. My friend can do this and has tried to teach me (i never really cared to practice it enough though) - its all about configuring the tongue and lips precisely over a clean tone. No distortion necessary. You only need distortion to create an undertone (really an artificial lower fundamental of sorts) for that real deep growly kind of throat singing with that name that begins with a K that I always forget how to spell.

[gno]

As I understand it, a formant is an overtone for a singer that can be modified by altering the vowel until it is properly reinforcing the pitch.

No - A formant is not an overtone. A formant is a resonant center. A sung or spoken pitch has harmonics (overtones) that are multiples of the fundamental. Those harmonics can be amplified by the resonant centers (formants) depending on how close the harmonics are to the formant.

A way you can hear the 2nd formant is by blowing air out your mouth (not a sung pitch) and form your mouth as if you were saying "ooooo weeee oooo". You can hear F2 rising and lowering in pitch as you change from "ee" to "oo".

[TimR]

Are you sure?

That's not the way I understood it

I don't want to say your wrong, because I'm pretty new to trying to learn vocal stuff, but I had some different ideas

My trombone does play overtones that are multiples of the fundamental; that is forced by the way overtones are made on a wind instrument; it also has a set of resonant centers that we call partials; some of these line up with the overtones and some do not; these resonant centers cannot line up because of the nature of the physical instrument with bends, straight parts, curved parts, flares, mouthpiece, leadpipe, etc

But voices? I thought the overtones were the formants and the overtones did not line up on multiples

I just sang a Bb4 into Audacity and checked the spectrum

Ah: 226, 470, 714, 915,1173

Eeh: 222, 471, , 916, 1162

Hmm, maybe they do line up as multiples, though the Ee is missing one

Better check the trombone:

221, 452, 687, 916, 1161

[Owen Korzec]

Pretty much all sounds create harmonics in multiples, the voice is definitely one of them.

The missing one in ee is because the formants are so spread that the middle harmonics aren't amplified.

Geno is correct about his description of formants. He is very well educated on the topic of formants.

[Owen Korzec]

What you are calling partials of a wind instrument is a very close equivalent to voice formants. The only difference is that vocal formants are adjustable - we can shift the dimensions of our instrument in many ways. So we can move those resonant centers around, and that is exactly how we form various vowels.

[TimR]

Pretty much all sounds create harmonics in multiples, the voice is definitely one of them.

.

In my world this is very definitely not true

Wind instruments, yes; but many others do not

(sorry the dog ate the period key on this laptop)

For example, piano, string instruments when plucked or strummed instead of bowed, all percussion including tuned ones like marimba, glockenspiel, temple blocks; handbells (I teach a handbell choir), etc None of these have overtones in multiples

But if you tell me the voice does, then I'm not going to argue, I'm here to learn, just that this is a new idea to me; I thought the vowel sounds meant the overtones were different for every different vowel on the same note

(but I can't sometimes resist asking a few question to be sure I understand, I'm not being argumentative)

Are you saying the overtones on a given pitch are always the same, but changing the vowel changes the relative strength of them by moving the resonances?

[Steven Fraser]

This is prompted by an argument over hearing overtones on wind instruments, but I thought I'd ask hear (here) <humor> about singer's overtones.

As I understand it, a formant is an overtone for a singer that can be modified by altering the vowel until it is properly reinforcing the pitch.

So it seems to me that what you listen for is the vowel sound, rather than hearing the formant as a distinct pitch.

I do hear a distinct overtone with throat singing, but not with normal voice.

So, can you hear the actual pitch of a formant? If so, is it worth learning to isolate them and hear them? Or does everybody listen for vowel, brightness, projection, etc., but not the overtones themselves?

Not sure I'm explaining this well, sorry.

Hi, TimR. Great question.

There is a tremendous amount on this topic available here and on the internet. Search for 'Source-filter theory' in ether place and you will find may very good resources to help you in your quest.

[Owen Korzec]

response in bold

In my world this is very definitely not true

Wind instruments, yes; but many others do not

(sorry the dog ate the period key on this laptop)

For example, piano, string instruments when plucked or strummed instead of bowed, all percussion including tuned ones like marimba, glockenspiel, temple blocks; handbells (I teach a handbell choir), etc None of these have overtones in multiples

really? where do you get this information? i am aware these all have very unique tones and may SOUND like they have overtones that don't fit, but I don't think that's the actual reality of what's happening.

For instance I'm certain that string instruments plucked still have harmonics in multiples, thus the ability to produce harmonics on the string with your finger - and those definitely go in multiples. There's no way I can think of that they would have funky harmonics outside of that. And because piano is a string instrument it applies to that too. Tuned percussion, not sure, they can definitely sound like they have odd overtones sometimes but this could be just due to which of the harmonics in multiples are amplified more than others. For instance if I hit an F on a marimba and I hear a pitch between D and D# along with it - that doesn't mean the harmonics are not multiples, it just means one of the higher harmonics (H7 in this case) is coming through stronger than the lower ones like the C an octave and a half above (H3) for instance.

And from what I can hear, temple blocks are pretty much just heavily dominated by the fundamental - and from what I researched, few sounds in real life actually truly have no harmonics - so a temple block would probably still have some, just less intense

But if you tell me the voice does, then I'm not going to argue, I'm here to learn, just that this is a new idea to me;

It does. Now there is a possibility to add non-harmonic "noise" into the voice in the form of airiness, undesirable tensions, or vocal distortion, but for a clean tone - I've looked at this on a spectrograph and so has anyone who here who has seen formants - the harmonics are all in multiples up to a point - anywhere between 3k-10k tends to be where the harmonics stop and the high end is taken over by "noise" which is usually just the tiny bit of airiness remaining in a balanced phonation.

I thought the vowel sounds meant the overtones were different for every different vowel on the same note

No, the overtones are static per pitch even as you change vowels but the INTENSITY of the overtones (which come from the source - the vocal folds) is varied BY changing the PITCH of the formants (resonances of the vocal tract that amplify said overtones). Does that make sense yet?

Do you happen to be familiar with how equalization (EQ) works? Vocal formants are essentially like EQ boosts for the voice. When we change vowels, it's essentially like having two EQ boosts and shifting their frequency to change the tone. You see EQ does not actually change the pitch of the harmonics of a signal at all (unless it's a very prominent and narrow EQ boost but this is not the case with vocal formants). The EQ just changes the intensity of the harmonics the source is already producing, by being adjusted to amplify the desirable ones.

(but I can't sometimes resist asking a few question to be sure I understand, I'm not being argumentative)

Are you saying the overtones on a given pitch are always the same, but changing the vowel changes the relative strength of them by moving the resonances?yes exactly. lol i totally just described that above before seeing this.

[Martin H]

When a harmonic produced by the glottal source is amplified by the resonance of the vocal tract it's called a formant. :)

[jonpall]

So is the answer yes or no?

[Owen Korzec]

To thread title? Yes and maybe, respectively

[MasterBlaster]

As I understand it, a formant is an overtone for a singer that can be modified by altering the vowel until it is properly reinforcing the pitch.

I would change what you said to "A formant is an overtone that can be modified by altering the vowel". YES!

So it seems to me that what you listen for is the vowel sound, rather than hearing the formant as a distinct pitch. YES!

So, can you hear the actual pitch of a formant?

The only isolable pitch that a "normal" person can hear when you listen to singing is the fundamental. YES you can "hear" the formant but you would have to be a musical "Rain Man" to be able to isolate it and determine it's distinct frequency without the aid of a software program. I don't have perfect pitch but if you were to sing "baaaaaaaaaaat" at A4 (440 hz) I would be able to figure out what note you're singing by comparing notes on my guitar until I found a match. And I could tell you if you were a little sharp or flat on the fundamental. But I would not be able to tell you exactly what your formant frequencies were without the aid of computer software. Scientifically I know that in addition to the fundamental A4 (440hz) you would be singing formant frequencies of aproximately 660hz (formant 1 ), 1720hz (formant 2 ) and 2410 hz (formant 3 ) to create the "baaaaaat" sound but there is no way in hell I could specifically isolate formant 2 and tell you you are actually singing it at say 1703 Hz.

If so, is it worth learning to isolate them and hear them?

NOOOOOOOO! Try singing at 2410 hz (D7 ish range). No point.

Or does everybody listen for vowel, brightness, projection, etc., but not the overtones themselves?

YES! The mouth , lips and larnyx positions will determine the formant overtones and thus the vowel sound and shape. The only pitch you need to "consciously" focus on is the fundamental frequency. Just worry about singing in tune and with clear articulate vowels sound and you are on your way. Oh by the way as you ascend in pitch you will need to modify your vowels here and there. But that's a whole other topic.

[MDEW]

Videohere keeps telling us to view vowels as throat shapes. This modifying of vowels is merely reshaping the vocal tract to enhance certain frequencies to project the voice better and find a better resonance space.

When you modify a vowel or tune a formant you are basically adjusting tongue, lip and pharynx positions for a better sound.

[TimR]

For instance I'm certain that string instruments plucked still have harmonics in multiples, thus the ability to produce harmonics on the string with your finger - and those definitely go in multiples. There's no way I can think of that they would have funky harmonics outside of that. And because piano is a string instrument it applies to that too. Tuned percussion, not sure, they can definitely sound like they have odd overtones sometimes but this could be just due to which of the harmonics in multiples are amplified more than others. For instance if I hit an F on a marimba and I hear a pitch between D and D# along with it - that doesn't mean the harmonics are not multiples, it just means one of the higher harmonics (H7 in this case) is coming through stronger than the lower ones like the C an octave and a half above (H3) for instance.

Okay. Any driven system will vibrate at the pitch(es) being input regardless of the inherent resonant frequencies of that system. However, it will only respond in a large way to those pitches near its resonant frequencies.

So when a clarinet reed puts a signal into an air column, or a violin bow puts a signal into a violin system, the system responds at the input frequencies. This is true even if you put a signal into a cast iron frying pan, say by touching a tuning fork, etc to it.

But that is for systems being driven by a signal input. A system that is struck with an impulsive force will then vibrate at its resonant frequencies until they damp out. Drop the frying pan on a concrete floor and you'll hear that happen. In engineering we commonly test systems by suspending them and tapping with a hammer, and using accelerometers to develop the spectrum. We also calculate them if the mass and stiffness matrices are amenable, and produce results called eigenvectors. I'm confident I'm right on this, this is fairly basic engineering and all my electives were in advanced vibration courses. Many mechanical systems fail prematurely if they are allowed to operate near resonances so we do calculate those when doing design. You really don't want your submarine propeller shaft coming apart at sea. I had one college who calculated the resonances of every screw in a machine tool and changed their sizes, which was taking it to a silly extreme, but he was kind of compulsive.

It makes sense that the voice would be a driven system, but I was confused by sources like the wiki article

http://en.wikipedia.org/wiki/Formant

that gives specific frequencies for different vowels, implying that singing that vowel will produce that frequency, and images that show that huge hump for the singer's formant that implies it is a relatively stable frequency overtone.

[Owen Korzec]

Okay so you're factoring in the additional resonance of what is striking the object. Okay that's fine and makes sense. Gotcha.

That being said, like you suspected, the voice is a driven system.

To answer your last question, formants are basically inaudible until they have a harmonic lined up near them. So if the harmonic is not tuned to the formant, the formant isn't heard.

So what this means is that singing "that vowel" will not produce the exact frequency written on the wikipedia page - it will simply amplify the frequencies of the harmonics of the pitch that are around the area of that frequency written on the wikipedia page.

So say you're singing a A3 at 220hz and the vowel is a kind of schwa "uh" with a F1 of around 500 hz and F2 of around 1500 hz. 500 hz will not be heard! Neither will 1500 hz! Because those frequencies do not coincide with the overtone series of a fundamental of 220hz.

BUT, what will instead happen is

Harmonics: 220 440 660 880 1100 1320 1540 1760 etc.

Formants: 500 1500

500 hz lines up closest to H2 which is 440 hz. So 440 hz will be boosted by the formant. and the harmonics above and below that - 220 and 660 might get a little boost too but not as much. Same with 1540 - the formant at 1500 hz lines up closest, so the intensity of 1540 hz will get boosted, and 1320 and 1760 will also get some boost but again not as much.

so what the ears will hear is something like

high intensity 440 1540

660 1320

220 1760

880 1100

low intensity

frequency on the horizontal axis, in hz

Now what happens with the singer's formant is that the harmonics up there are so close together that the formant can be static and will still pick up and amplify the higher harmonics of whatever pitch it's given.

For instance let's take a G4 an A4 and an C5

let's say the singer's formant is at 3000 hz.

(rounding fundamentals of G and C to nearest hundred for ease)

G4 harmonics: 400 800 1200 1600 2000 2400 2800 3200

A4 harmonics: 440 880 1320 1760 2200 2640 3080

C5 harmonics: 500 1000 1500 2000 2500 3000

the singer's formant will easily amplify the 3080 hz of A4 and the 3000 hz of C5 and since it's a cluster it is actually a bit wide (e.g. may have one formants at 2800 hz, one at 3000 hz, one at 3200 hz) and will even amplify BOTh the 2800 hz and 3200 hz of the G4.

Pretty much any pitch you sing the singer's formant will amplify.

However, if I tried to keep a static F1 of 1000hz on an ah vowel for instance, look what happens if I try to sing a G4 (again rounding here)

G4 harmonics:400 800 1200 1600 2000

F1 freuqnecy: 1000

the formant lands in between two harmonics, which is the same as the singer's formant over that G4, but since F1 is only one formant, not a cluster of F3/F4/F5 and the pitches of 800 hz and 1200 hz are waaay wider (a half an octave apart!), what will realistically happen is neither the 800 hz H2 nor the 1200 hz H3 will get amplified much. and so the singer has to either tune to F2 instead, rely on the singer's formant cluster, or, what people usually do is modify the vowel (for instance an AW may help bring down F1 to 800 hz so it can line up with the 2nd harmonic) so that the formant lines up better to one harmonic - so there is more intensity in the lower frequencies, perceived as a boomier tone instead of just relying on the singer's formant frequencies which would "cut" through a mix fine, but be perceived as a thin tone if not supplemented with some nice resonance in F1 or F2 - one of the vowel formants.

I know this is a lot but you seemed very confused so I hope this just made it very clear.