The sounds we voice depend on how we manipulate the tube defined by the mouth and throat. When making a sound like eee, for example, the mouth constricts but the throat opens. In a sound like aah, the mouth opens and the throat constricts. To study precisely how we make sounds, speech scientist Brad Story of the University of Iowa lay in a magnetic resonance scanner for three seven- to eight-hour sessions. I don’t think I would do it again, he says. Story made 22 different vowel and consonant sounds while the scanner recorded the shape of his vocal tract and then processed the images to represent the tract as a solid tube, dissected from the skull. When he measured the cross-sectional areas of the tube for each sound and fed the information into a speech simulator, the voice it produced sounded remarkably like his own. Story next hopes to create a virtual vocal tract on a computer to study the nuances of the voice. We might learn how singers can enhance their performance, or how a cleft palate affects the voice, says Story. In speech it’s very difficult to get at a lot of things that we want to understand, because humans aren’t willing to have needles and probes stuck through their vocal tracts.