Voices of the Past
The sound of a vocal tract from the past has been synthesized to be heard again in the present. The Egyptian Nesyamun lived during the politically volatile reign of pharaoh Ramses XI (c.1099–1069 BC) over 3000 years ago, working as a scribe and priest at the state temple of Karnak in Thebes (modern Luxor). The restoration of an exact vocal sound requires the perfect preservation of the soft tissues, which is impossible for individuals whose remains are only skeletal. Even where soft tissue does survive, for example in mummified remains, the vocal tract can either be missing or distorted. The process is only feasible when the relevant soft tissue is reasonably intact, as in the case of the 3,000-year-old mummified body of the Egyptian priest Nesyamun.
Following the unwrapping of his body in 1824, it was examined by members of the Leeds Philosophical and Literary Society including three surgeons and a chemist. The multidisciplinary scientific investigation published in 1828 was first of its kind. Following the development of X-rays, the body underwent a radiological examination in 1931 at the University of Leeds’ School of Medicine, in 1964 by the University of Sheffield School of Dentistry, and in 1990 at the University of Manchester by a team using endoscopy, histology, X-ray and early CT scanning techniques. These combined studies revealed that Nesyamun had died in his mid-50s and had suffered from gum disease and severely worn teeth.
In September 2016 Nesyamun’s mummified body was transferred from Leeds City Museum to the nearby Computed Tomography (CT) Scanning Department at Leeds General Infirmary. The CT images confirmed that a significant part of the structure of Nesyamun’s larynx and throat remains in situ as a result of the elaborate mummification process, thus enabling the vocal tract shape to be measured. The tongue, however, has lost its muscular bulk over time and the soft palate is not present.
Following the scans, a 3-D printed tract was created for Nesyamun and designed to be used with the Vocal Tract Organ which provides an appropriate acoustic larynx source as a time-domain waveform synthesis of the Liljencrants-Fant (LF) larynx source which is commonly employed in speech synthesis.
ITK-SNAP, which allows a three-dimensional structural representation of human tissues to be observed, was used to view the airway between the larynx and lips which is itself isolated as a solid shape to enable the 3-D printing process. A virtual sheath is created around the airway to which a loudspeaker coupler is added. The resulting vocal tract model is 3-D printed.
The ‘in death’ vocal tract acoustic output of Nesyamun has been scientifically synthesised. This acoustic output is for the single sound for the extant vocal tract shape; it does not provide a basis for synthesising running speech. To do so would require knowledge of the relevant vocal tract articulations, phonetics and timing patterns of his language. The synthesised vowel sound based on the precise dimensions of his unique vocal tract is here compared to modern vowels as proof of method and to demonstrate future research potential.