The human voice is a wind instrument too. The vocal folds, in the larynx, produce a sound rich in harmonics; the vocal tract (throat, mouth and lips) is a tube of variable shape, about 17 cm long in an adult, whose resonances reinforce some regions of the spectrum: the formants. By changing the position of the tongue, the jaw and the lips we change the shape of the tube and therefore the formants, and this is how we pronounce different vowels. In these experiments with TubeLabLight we use the vocal-tract profiles of seven Italian vowels.

The Italian vowels

1. Instructions
   • Open TubeLabLight with the /a/ (casa) project. The drawing shows the profile of the vocal tract: on the left the glottis (where the vocal folds are), on the right the lips. In place of the mouthpiece choice there is the Voice pitch slider.
   • Press ▶ Listen: you hear a voice pronouncing a sustained "a".
   • Read in the table of modes the frequencies of the first two modes, M1 and M2: they are the first and second formants, F1 and F2. Here the notes column is of no interest.
   • Repeat with all the vowel projects: /i/ (vino), /e/ (avere), /ɛ/ (bello), /a/ (casa), /ɔ/ (cosa), /o/ (forma), /u/ (uva) (the Italian words show how each vowel is pronounced), and write F1 and F2 for each one in a table. If the sound is on, changing project changes the vowel without interrupting the voice.
2. Remarks
   • You should find values close to these:

     vowel	/i/	/e/	/ɛ/	/a/	/ɔ/	/o/	/u/
     F1 (Hz)	278	361	565	828	528	429	283
     F2 (Hz)	2271	2066	1883	1294	905	808	727

   • The first formant increases going from /i/ to /a/ and decreases again going from /a/ to /u/: it is related to how open the mouth is, and it is largest for the most open vowel, /a/.
   • The second formant is high for the "front" vowels /i/, /e/, /ɛ/, in which the tongue rises towards the palate (in the drawing, the narrowing in the right half of the tube), and low for the "back" vowels /ɔ/, /o/, /u/, pronounced with the tongue back and the lips rounded.
   • In the The vowel triangle chart, below the table of modes, the app plots the points (F1, F2) of the seven vowels and, with a cross, that of the current tube: the vowels lie on a triangle with corners at /i/, /a/ and /u/.
3. Related topics: see Vowels, Simple models of the voice, The vowel A
4. To find out more: see Perception of timbre, How to read a spectrogram

Changing the note without changing the vowel

1. Instructions
   • Load the /a/ (casa) project and press ▶ Listen.
   • Slowly move the Voice pitch slider from 90 to 250 Hz and listen. Meanwhile, look at the green curve in the chart and at the formants in the table of modes.
   • Repeat with the vowel /i/ (vino).
2. Remarks
   • The note rises, but the vowel stays an "a": the formants depend on the shape of the vocal tract, not on the frequency of the vocal folds, and indeed the green curve and the table do not change. The harmonics of the voice move and spread apart as the note rises, but their amplitude always follows the profile of the green curve (apart from a slow decrease towards high frequencies, due to the source): the envelope of the spectrum stays the same.
   • When the note is high the harmonics are further apart and "sample" the formants at only a few points. In singing, on very high notes, this makes the vowels less and less distinguishable.
3. Related topics: see Vowels in singing, Perception of timbre, Fourier's theorem
4. To find out more: see Physiology of the auditory system, Perception of pitch

The length of the vocal tract

1. Instructions
   • Load the /a/ (casa) project and note the first two formants.
   • Press the +5% button next to Length three times: the vocal tract gets longer, from 170 to about 197 mm. Note the formants again and listen.
   • Reload the project and press −5% three times: the vocal tract gets shorter, about 146 mm.
2. Remarks
   • Lengthening the vocal tract lowers the formants (F1 from 828 to 724 Hz, F2 from 1294 to 1120 Hz); shortening it raises them (F1 to 951 Hz, F2 to 1504 Hz). As in any tube, the resonance frequencies are inversely proportional to the length.
   • The vowel is still recognisable, but the voice seems to belong to a larger or smaller person. A woman's vocal tract is on average shorter than a man's, and a child's shorter still: the same vowels have higher formants. This is why, when comparing the formants of different people, their relative position matters more than their absolute value.
3. Related topics: see Simple models of the voice, Natural frequencies of pipes

Rounding the lips

1. Instructions
   • Load the /a/ (casa) project and press ▶ Listen.
   • Press Round the lips: the radius of the last two points (x = 164.9 and 170 mm) becomes 6 mm. It is like narrowing the lips, whose opening goes from a radius of about 21 mm to 6 mm. Listen and read the formants.
2. Remarks
   • The formants drop, F1 from 828 to 476 Hz and F2 from 1294 to 1216 Hz, and the vowel "darkens" towards an /o/. Narrowing the outlet of a tube lowers its resonances, as happens in a Helmholtz resonator when its neck is narrowed.
   • This is what we really do when we pronounce the rounded vowels /o/ and /u/: narrowed, protruding lips lower the formants (in the /o/ and /u/ profiles the last stretch of the tube is narrow).
3. Related topics: see Vowels, Helmholtz resonators, Simple models of the voice
4. Other experiments
   • Start from /i/ and try to turn it gradually into /u/ by moving the points of the profile with the mouse, watching how F1 and F2 change.