Twenty Years Before Edison’s First Phonograph


It turns out that nearly twenty years before Thomas Edison produced what is usually thought to be the first voice recording (“Mary Had a Little Lamb”, in 1877), Édouard-Léon Scott de Martinville, a Parisian typesetter and tinkerer, produced a ten-second audio snippet that was recently discovered.

In 1860, de Martinville used his phonautograph to produce a ten-second recording of a woman singing “Au Claire de la Lune.” (Here is a clearer version of the same song, 1931). Apparently, at the time, a technique existed for recording the sound (however coarsely), but none for playing it back (sort of like suspended animation today—we can freeze you fine, but we can’t defrost you). 

The user spoke into a large barrel, and the sound waves were funneled down to a small stylus.  The stylus made squiggles on a piece of rag paper that was covered with soot from an oil lamp. Contemporary researchers at Lawrence Berkeley National Laboratory were able to scan this image in digitally and use algorithms to recover the sound from it. 

Here’s the full story.


Here’s excepts of another story on the audio recording being played for 150 scientists, musicologists, audiophiles and phonograph collector at a Stanford recital hall. 


Physicists convert first known sound recording

David Perlman, Chronicle Science Editor


The voice of an unknown woman singing in a lamp-lit Paris laboratory nearly 150 years ago came to life Friday amid the crackles and buzz of a historic breakthrough recording made 17 years before Edison invented the phonograph.

“It’s ghostly. It’s magic,” audio historian David Giovannoni said of the sounds that filled a Stanford recital hall. “This voice is a young woman trying to come into the 21st century to sing for us.”

Giovannoni played the sound again for a loudly applauding audience of 150 scientists, musicologists, audiophiles and phonograph collectors who had come to hear the long-ago French soprano singing “Au clair de la lune” in warbling tones restored by physicists at the Lawrence Berkeley National Laboratory.

Instead, he said, Scott meant to preserve in graphic form the great music and “declamations” of the day in a form that could be read like writing far into the future.

Modern acoustic science, however, can do much more.

The voice singing about Pierrot is believed to be Scott’s own daughter. The year after it was recorded, Scott turned over the blackened sheet of paper to the French Academy of Sciences, Giovannoni said.

The two Berkeley physicists were helped in recovering the “Pierrot” voice immeasurably, Cornell said, by the fact that Scott had recorded the sound waves together with waveforms from a tuning fork vibrating at a constant frequency of the musical note A above middle C. That enabled the scientists to compensate for the fact that Scott’s device – like Edison’s first phonographs much later – was hand-cranked and wobbly.