The Story of the First Recorded Human Voice

By Wesley Fenlon

MacArthur Fellow Carl Haber was instrumental in playing back a French recording made in 1860.

Physicist Carl Haber is one of 2013's MacArthur Fellows, the recipient of a genius grant from the MacArthur Foundation. Haber deserves the genius label. Over the past five years, his work on a system called IRENE at the Lawrence Berkeley National Laboratory has helped play back audio recorded more than 100 years ago, even when the recording medium was obsolete or in terrible condition. If that concept sounds vaguely familiar, it should--earlier this year Haber was the first person to hear Alexander Graham Bell's voice preserved in an 1885 recording.

As described by the MacArthur Foundation, IRENE (which stands for Image, Reconstruct, Erase, Noise, Etc.) is "a non-contact method for extracting high-quality sound from degrading or even broken analog recordings on two- or three-dimensional media. A disc or cylinder is placed in a precision optical metrology system, where a camera following the path of the grooves on the object takes thousands of images that are then cleaned to compensate for physical damage; the resulting data are mathematically interpolated to determine how a stylus would course through the undulations, and the stylus motion is converted into a standard digital sound file."

If you try to play a vinyl record on your average turntable, a lot of things can wreck your sound. Scratches create annoying skips. Dust collecting on the needle can muddle playback. And if a record is broken, it's unplayable. But that's not the case with IRENE, which can create a digital image of a record even if it's broken into pieces. IRENE is smart enough to skip over scratches and cracks and, essentially, create a musical photocopy.

The voice of Alexander Graham Bell holds a special place in the history of technology, since his invention of the telephone revolutionized communication. Bell's recordings weren't the first that Haber helped restore, though. Back in 2008, he helped play back the earliest known recording of a human voice, which was made in 1860--17 years before the invention of the phonograph, and 25 years before Bell's recording. In this case, "recording" doesn't mean exactly what we would think.

French printer Édouard-Léon Scott de Martinville invented phonautograms in the 1850s and was interested in recording human speech--playing that speech back like a record, however, was not part of the plan. Wikipedia explains that Scott's phonautograph "transcribed sound waves as undulations or other deviations in a line traced on smoke-blackened paper or glass. Intended solely as a laboratory instrument for the study of acoustics, it could be used to visually study and measure the amplitude envelopes and waveforms of speech and other sounds, or to determine the frequency of a given musical pitch by comparison with a simultaneously recorded reference frequency."

Scott's sound recordings are archived on the website FirstSounds.org, which chronicles the effort that went into playing them back. For a century and a half, playing them seemed impossible. Then, in 2008, came IRENE. Haber's technology was able to read the lines scratched in soot as though they were the grooves of a record, creating a digital file.

Playing them back, however, was still a huge challenge.

FirstSounds writes "From a modern perspective [Scott's] tracings are often 'malformed': the recording stylus sometimes left the paper and sometimes moved backwards along the time axis, violating basic assumptions of the 'virtual stylus' approach and—for that matter—of sound recording in general. For this reason, we supposed at first that many of Scott’s phonautograms—particularly the earliest ones—might remain permanently mute."

Then one of FirstSounds' cofounders created an alternate playback approach "graphically converting phonautographic wavy lines into the edges of bands of variable width and playing these back using software designed to handle optical film sound track formats." When that method was applied to Scott's 1857 recordings, they produced sound, but not especially clear sound. But with Scott's 1860 recordings, FirstSounds struck gold.

Scott added 250Hz tuning fork calibration traces to his phonautograms, which gave FirstSounds a timecode to base their recreation on (since these recordings were created with a hand crank, there was no consistent recording speed). The resulting recording isn't nearly as impressive as Alexander Graham Bell's clear voice, but it's still incredible that the collaborations of Haber and other scientists were able to produce an audible archive of something recorded on paper in 1860.

If anyone had recognized the incredible potential of the phonautogram at the time, they could've recorded the Gettysburg Address, which Lincoln delivered in 1863. Recordings of French songs will have to do.