14: The Human Voice

As we know, musical instruments consist of a vibration which is amplified by resonance. The human singing voice is no different. The vocal chords are the vibrating part and the throat, mouth, nasal cavities and bronchial tubes constitute the resonance cavities that amplify these vibrations into sound. Because every person's combination of throat, mouth, nasal cavities and bronchial tubes is slightly different, we all sound slightly different. The fact that we can change the shape of some of these cavities at will enables us to produce a wide range of pitches, depending on the initial structure and training. Here is Wikipedia on the human voice.

Key Terms:

Vocal cords, vocal tract, Bernoulli's principle, resonance cavities, vocal formants, trachea, larynx, phonemes (plosive, fricative, vowels, diphthongs, consonants, semivowels, gliding consonants, liquids, nasal), throat singing, articulation (opera singers).

14a: The Vocal Tract

The main components of the vocal tract are sketched in the figure below. The nasal cavity (1) is actually a set of connecting chambers in the front part of the face, above the oral cavity (2) or mouth. The hard palate (3) separates the mouth from the nasal cavity. The soft palate (4) extends the hard palate and connects with the uvula (6), the part you see hanging down in your mouth when you look in the mirror. The teeth (5), lips (7) and various parts of the tongue (9, 11, 13, 15) are also shown. The pharynx (8) connects the mouth to the trachea (16) or wind pipe and the esophagus (not labeled). The epiglottis (10) closes the trachea when you swallow food so that the food goes down the esophagus and not into your lungs. The vocal cords (12) or folds are located at the glottis (14), the external part of which is called the larynx (17).

The vocal cords or vocal folds are actually two folds of skin that can be brought together to partially block the flow of air coming from the lungs. When they do this the Bernoulli effect causes them to snap shut until the air behind them builds up enough force to them open again. The picture on the left is a picture of the folds (looking down into the throat from the mouth) when the folds are relaxed and open. On the right is a diagram of the vibration process.

In the same way that a brass player's lips buzz, the vocal cords open and close to form a buzzing sound. By tightening the muscles around the flaps we can change the frequency of the buzz but only slightly. The following are videos of vibrating vocal cords in singers.

Video/audio examples:

Video with explanation of vocal cords while singing.
Web page with several videos of vocal cords while singing.
YouTube of Mel Blanc's vocal cords.

14b: Vocal Formants

As in the case of a brass instrument, the shape and size of the resonance cavities in the body select some of the buzzing frequencies but diminish other ranges of frequencies. The nasal cavity, the oral cavity, the larynx, the trachea and the lungs are all air cavities that can have an effect on the sound being produced. Lets look at a simple example; buzzing sound fed through three Helmholtz resonators.

The buzzing sound produced by the vocal folds has a wide range of frequencies between 60 Hz and about 7,000 Hz when overtones are included (technically it is called pink noise which will be discussed in Chapter 18). Below is a diagram of the frequencies (in hertz) of pink noise; all frequencies are present but the amplitude gets progressively smaller for higher and higher frequencies.

Suppose we have three Helmholtz resonance cavities, all linked together as shown in this figure:

These resonating cavities are of three different sizes so they will have three different resonance peaks (review Chapter 4 if you have forgotten what resonance is) as shown in the following figure.

The result of feeding pink noise into the three different resonators is shown in the graph below (frequency in hertz). Notice that the available frequencies are now limited by an envelope as the result of the resonance cavities. This envelope imposed on the spectrum of available frequencies is called a formant. Its particular shape is controlled by the shape of the resonance cavity.

A simple model of the human vocal system is that of two resonance cavities, the mouth and the larynx. Both cavities change shape when you talk or sing but the mouth changes more. In the following figure the vocal tract is modeled for the sound 'ah' (written as /a/ in the International Phonetic Alphabet). If you think about what shape your vocal tract has when it makes this sound, the mouth is very open. Below the vocal tract figure is a two cavity model with the first cavity (the mouth) bigger than the second. A sketch of the formant is also shown on the right. Notice that lower frequencies are emphasized over higher frequencies.

When you make the sound "ee" (written as /i/ in the International Phonetic Alphabet) the mouth closes up to make a small resonance cavity. The figure below shows the vocal tract, a two cavity model and the approximate formant. For this sound there is more sound at higher frequencies as seen by the three peaks on the right side of the graph.

Here is a graph of formant vowel frequencies. The graph shows the first two formants of English vowel sounds. Notice that the /i/ sound ("ee" as in the word 'seat') for most speakers lies in the upper left of the graph so the first formant has low frequencies and the second formant has high frequencies. The /a/ ("ah" as in the word 'sought') for most speakers lies towards the lower right; high first formant frequencies and second formant frequencies lower than for /i/.

Because the mouth cavity and larynx are so flexible, a wide variety of formants can be made resulting in a wide variety of sounds. Try saying the words "heed", "hid", "head", "had", "hod", "hawed", "hood", "who'd". How does your mouth and larynx structure change for each sound?

Video/audio examples:

Here is an interactive simulation of formant vowel sounds. Click 'Launch' and you can manipulate the shape of a human vocal tract and hear the resulting changes in sound.
Here is a video of a talking mouth, created by scientists at Kagawa University, Japan. You will notice something is missing in the sounds. This is because the robot only produces formants; it does not have a tongue or lips to produce the other phonemes needed for speech.
Here is an interactive graph of formant vowel frequencies.
The vowel formants, in Canadian English.
Formants, explained with sample frequencies.
Formants, explained with sample frequencies and MRI images.

14c: Phonemes

The distinct sounds that make words spoken in a particular language are called phonemes. Different languages often use distinct phonemes. We learn how to make these distinctive sounds very early in life by imitating the sounds around us. People who learn to speak a different language as adults often have an accent because they are using the phonemes they learned for their first language instead of the correct phonemes used by a native speaker. Phonemes in English can be grouped into five categories; plosives, fricatives, pure vowels, diphthongs and gliding consonants. A sixth category involving nasal sounds is sometimes included in this list.

As mentioned in the previous section, frequencies emphasized by the vocal formants play a role in each type of phoneme. In much the same way that attack frequencies for a musical instrument affect the identification of a musical sound by the ear-brain system, how the formant frequencies are initiated and terminated affect the way they are heard. For example plosive consonants are formed when the air flow through the vocal system stops completely. Examples in English would include the letters 'p', 't', and 'b'. The sounds 'g' as in 'gun' and 'k' as in 'kick' are also included as plosives. Changing the way the air stops or starts can affect the transient frequencies at the beginning and end and this changes the perceived sound. For example the sounds of the letters 'p' and 'b' have basically the same formants but different initial plosives.

Fricatives (sometimes also called sibilants) are another variation in sound in addition to the formant that has to do with the way it is produced. Examples are 'f', 'v', 'th' (as in 'thin'), 'th' (as in 'these'), 's', 'z', 'sh', and 'zh' (as in 'measure'). These sounds include frequencies that are caused by air turbulence in the mouth, in addition to the frequencies created by the vocal cords.

Most vowels are phonemes created by the action of formants on the vocal cord frequencies. For some vowels or vowel combination and for some consonants the sound changes while being made. In the case of diphthongs, two sounds are put together with a glide connecting them. Examples in English are found in the words 'eye', 'hay', 'boy', 'low', and 'cow'. A similar sort of thing happens for the gliding consonants which can further be broken down into semivowels such as 'w' and 'y' and liquids such as 'l' and 'r'. In all of these cases the formants change during the creation of the sound.

A final category of phonemes are the nasal sounds. For these phonemes the nasal cavities are used to change the vocal formants to produce sounds such as 'm', 'n' and 'ng' (as in English).

The phrase "SAY BITE AGAIN" has several phonemes. "S" is fricative "AY" is a diphthong, "B" is a plosive"I" is vowel, "T" is a plosive, "AG" is a plosive, "AY" is a diphthong and "N" is nasal. Here is a sound file of "SAY BITE AGAIN" spoken normally.