Creating a voice: a challenging inverse problem (part 2)

(a continuation of this previous post)

In this post, I will describe how the human voice is produced and why this is useful to the approach of the Inverse Problems Group at University of Helsinki.

Our voice is produced by air flowing from our lungs through our vocal folds and vocal tract. Vocal folds - commonly known as vocal cords - are a part of the glottis and flap to produce a sound from the flowing air. They move extremely fast (hundreds of times per second) and "break" the air flow creating sound (*). The dimension of vocal folds characterises the pitch: women's vocal folds are usually smaller than men's. The pitch is related to the number of vibrations per second, as you can see in the following chart:

Link between music notes and movements per second of the vocal fold (courtesy of
Link between music notes and movements per second of the vocal fold (courtesy of

Here you can watch a stroboscopic video of vocal folds moving (the slow motion is an effect of the stroboscopic light):

Another characterisation of the voice pitch comes from the vocal tract, resulting to be generally longer in men. The vocal tract is quickly deformed while we talk, as you can notice in this real-time MRI video:

The vocal tract acts as a resonator and its shape changes the "buzzing sound" produced by the vocal folds into a voice sound as we hear it. For studying purposes, our group printed some approximation models of the vocal tract corresponding to five vowels:


In the next post I will (finally!) unveil how inverse problems meet human speech. Wait for it!

(*) To know more about this topic, check out "Principles of voice production" by I. R. Titze.

Share on FacebookShare on Google+Share on LinkedInTweet about this on TwitterEmail this to someonePrint this page

Paola Elefante

Technical Project Manager working in Supply Chain Management solutions at Relex Solutions Oy. Proud mother with the best husband ever. Shameless nerd&geek. Feminist. Undercover gourmet.

One thought on “Creating a voice: a challenging inverse problem (part 2)

Leave a Reply

Your email address will not be published. Required fields are marked *