Scientists recreate a mummy's voice
Scientists have managed to recreate the lost voice of a 3000-year-old mummy, Nesyamun - a high ranking priest in ancient Egypt. Using an MRI scanner, they imaged the structure of his throat and recreated it with a 3D printer. And by using a speaker to simulate the vibrations that would come from moving vocal cords, Nesyamun can speak and sing again - ish. Adam Murphy heard from Royal Holloway, University of London’s David Howard how you go from mummy to maestro…
David - You’ve got to get him to a hospital. And that’s clearly done out of hours, for fairly obvious reasons. He then has to be lifted out of the sarcophagus onto the MRI table, to be taken into the machine. The nice thing is of course, he doesn’t move. For Nesyamun the pictures are very clean indeed because of course he’s not breathing, so nothing is moving. You then take a scan of the area from below the larynx in the neck to the upper end of the nasal cavity so that we’ve got all the vocal apparatus. And then you, in the computer, identify the airway in that, which comes out in contrast to soft tissue and bone. We then extract that from the airway in software, and you can then put a sheath round it in the virtual world, so a virtual sheath goes round it that’s 2mm thick. And finally you extract the sheath itself and send it to the 3D printer, which then produces the outline of the actual vocal tract that’s in his head and neck. That is placed on the loudspeaker with a special little coupler we’ve got designed that fits over the top, and we then put the sound in and listen to the result.
Adam - And how close is the recreation to how Nesyamun would have actually sounded?
David - We believe acoustically the sound that you’re hearing is exactly the sound that he would make if he were suddenly to become alive again and just make his vocal folds vibrate. And we’ve done work on that in the past looking at living people, including myself, where we make scans and we compare them acoustically with the original speaker. So we are confident that what we’re hearing is how he would sound if he spoke exactly as he is in his sarcophagus; bearing in mind that his tongue has lost its main muscular bulk, so that area of his tract is not as it would have been when he was alive and speaking.
Adam - Amazing. And is there any way potentially to get speech out of this, to get a three-thousand-year-old voice talking?
David - We’ve been thinking about this. At the moment you can’t do that with a plastic vocal tract because you can’t move it around, it's solid. But if we were able to calculate the sound coming out of the lips in the computer, for a larynx input, and we were also able to move elements of the vocal tract in the computer while it’s calculating it, then you’d have the basis of doing what is known as articulatory synthesis. And elements of that can be done. So I think it is plausible to think that we could make him articulate sounds. The first step for that would be to reinstate the missing part of his tongue, so that that tongue muscular bulk is there to allow sounds to be made, different vowel sounds and so on.
Adam - And then just lastly: what do you think of his voice? You have some musical experience behind you, do you think he’s got a chance on the Voice?
David - Yeah, I think he probably has. I mean, he’s got the right acoustic characteristics as is shown in the paper, so he certainly has that side of it. But what we do know is that he did lead the worship and he did indeed sing. And he did this regularly, because he was a priest to Rameses the Eleventh, so that was part of his day job. And my Egyptian colleagues tell me that the actual material he sang is documented: how it was pronounced, and also in terms of the music to which he sang. One of my big interests here would be to try and recreate some elements of his sung output.