Phonetic Exercises

Phonetic Exercises - Languages in Danger

Author: Maciej Karpiński.
Amazing voices

Overtone singing

Overtone singing is a difficult technique that enables singers to produce overwhelming, powerful, almost unhuman sounds. A famous Huun Huur Tu group of singers coming from Tuva contributed to making it world-wide known:

With its ca. 265.000 speakers, Tuvan is not endangered but still does belong to smaller and less studied languages.

For our analyses, open a different recording by the same group:

It starts with an a capella solo. Run Visual Analyser which is able to show the spectrum in real time. If you run it on the same computer as the one used for playing the music, some interferences may occur. If this is the case, transfer the song to any other music-player that would retain relatively high quality (e.g., your mobile music player, or a mobile phone) and then play it back into the microphone of your computer, running Visual Analyser.

Watch closely the shape of the pulsing line in the bottom panel. Its shape represents the energy of voice for various frequencies. There are many little peaks. But you can spot some more prominent ones. Where? When do they occur? Can you count them?

[New peaks occur when new overtones are added by the singer. Of course, they are present also in regular speech and singing but here the singer manipulates them consciously using this peculiar singing technique. Singers can generate one or two overtones which are perceived as somehow independent from the basic tone (drone).]

For the same portion of the song, watch the upper panel where the oscilogram is presented. Watch closely the shapes it takes. Stop the player and then try to produce “regular” sounds of similar pitch. How is your oscilogram different? Can you remember the characteristics of spectrum for the Tuva singer's voice? How is your voice different?

You can use the “zoom” function of Praat (you’ll find respective buttons in the left lower corner of the “analysis window”). Magnify the signal to the limits. You’ll see that the oscillogram is actually built of separate points, corresponding to measurements made in equal intervals by the mechanism of analogue-to-digital convertion.

[Of course, much depends on how you produce your sound and what the quality of your voice is. With simple, “round and clean” sounds, you will see a sinusoid-like shape. When you add some harsh quality, small distortions will occur.]

It's just amazing!

As a reward for your efforts, you can listen to this incredible performance: