What are formants?

Formants are the frequency regions where the vocal tract resonates and strengthens the voice. Their positions, mainly the first two, are what makes an “ee” different from an “ah”.

Formants and harmonics are not the same thing

A sung note is made of a fundamental frequency (F0) and its harmonics at whole-number multiples of it. These come from the vocal folds. Formants belong to the vocal tract: they are broad peaks in the spectrum that show which harmonics get boosted. Changing the pitch moves the harmonics but not the formants; changing the shape of the mouth moves the formants but not the pitch.

A simple tube model

A neutral vocal tract can be pictured as a straight tube, closed at the glottis and open at the lips. Such a tube resonates at odd multiples of c/4L, where c is the speed of sound (about 350 m/s in warm, moist air) and L the length. For L = 17.5 cm this gives about 500, 1500, 2500 and 3500 Hz, the neutral formants F1–F4 of an adult male. Making the tube longer lowers them; making it shorter raises them.

Typical values for vowels

F1 follows how open the mouth is and F2 follows how far forward the tongue is. For adult male speakers the averages are roughly: “ee” (as in see) F1 270 Hz, F2 2300 Hz; “ah” (as in father) F1 730 Hz, F2 1100 Hz; “oo” (as in boot) F1 300 Hz, F2 870 Hz. Women's values are about 15–20 % higher because the tract is shorter, and individual speakers vary.

Formants in singing

When a singer's pitch rises above the first formant of a vowel, the voice loses the boost that F1 gave it. Many sopranos therefore open the jaw wider on high notes so that F1 follows the pitch. At the same time, the third to fifth formants can be brought together to make the singer's formant, which helps a voice to be heard over an orchestra.

Measuring formants

Formants are usually estimated from a recording by linear predictive coding (LPC) or read from the spectral envelope. The estimate is reliable only when the harmonics are close together: at high pitches there are few harmonics under each peak, and the result can be wrong. The Voice Lab's tuner shows its confidence in the formant reading for this reason.

Questions and answers

What do F1 and F2 mean?

They are the first and second resonances of the vocal tract. F1 mostly reflects how open the mouth is (jaw height), F2 how far forward or back the tongue is. Together they identify most vowels.

How many formants does a voice have?

In principle a tract has many; in practice the first three to five matter for speech and singing. F1 and F2 identify the vowel, higher ones affect the voice quality.

Why are formants hard to measure in high voices?

Because the harmonics are far apart, so the spectrum samples the resonance curve only at a few points and the peaks cannot be located reliably.

Go deeper: lessons

Open the 3D model