Until this week’s announcement, the earliest known sound recording was Edison’s 1877 “mary had a little lamb” experiment. Today, we heard a woman’s voice recorded in 1860. We now know that a Parisian named Ã‰douard-LÃ©on Scott de Martinville made an audio recording 17 years before Edison (actually, Martinville attempted audio recordings as early as 1853, but they are unintelligible).
This discovery does not diminish Edison – Martinville had no idea how to play back his phonautographs.
The 1860 recording was discovered in an obscure French archive earlier this month by a team of American researchers. It’s quite the detective story. The French phonautograph was extracted via optical techniques developed by Lawrence Berkeley Labs. Raw, semi-processed, and fully-processed results were played publicly today for the first time. A semi-processed MP3 version is on-line here.
What’s really fascinating is that Martinville’s method recorded TWO adjacent tracks. One track was the female voice program while the other track was a tuning fork specified by Martinville as 435 Hz. This of course allowed Berkeley researchers to lock absolute pitch. Brilliant!
This afternoon, I spoke at length with the project’s lead researcher (David Giovannoni) who shared, “we’ve found evidence of even earlier Martinville recordings of similar caliber.” This remarkable story is likely not finished.