Tutorials Audio

Édouard-Léon Scott de Martinville: Writing Sound Down, 17 Years Early

By Infinite Creation · Profile

Historical photograph of Édouard-Léon Scott de Martinville.

At a Glance

Édouard-Léon Scott de Martinville (1817-1879) built the phonautograph in the 1850s, a device that traced airborne sound as a visible line on lampblacked paper. He was making a written record of sound for study, not a reproducing machine, and it appears never to have occurred to him to play the traces back. In 2008 researchers scanned surviving phonautograms and recovered the audio, including a voice singing "Au clair de la lune" from 1860, seventeen years before Edison.

Lived: 1817-1879, France
Anchor year: 1860, the earliest recovered recording
Strand: overlooked, recording without playback

Career

The problem as it stood

Sound was studied by ear and by analogy. There was no way to capture a waveform for examination, and the science of acoustics had no equivalent to the photograph.

What he actually did

Scott built a barrel-shaped horn terminating in a membrane with a stiff bristle attached. Sound vibrated the membrane, the bristle scratched a trace into soot on a rotating cylinder of paper, and the result was a written record of the sound. He patented it in 1857 and made recordings through the early 1860s.

How it worked

It is a phonograph missing exactly one idea. The transduction, air pressure to mechanical displacement to an inscribed groove, is the same. What Scott lacked was the inverse step: running the trace back past a stylus to recreate the vibration. Edison's insight was not how to record but that the recording could drive the process backwards.

What it made possible

At the time, the study of vowel waveforms and little else. In retrospect it establishes that recording and reproduction are genuinely separate inventions: the single most useful distinction for reading the rest of this history.

What happened to him

He watched Edison receive credit for sound recording and protested that the essential step had been his, without effect. He died in 1879. In 2008 the First Sounds initiative used optical scanning to convert his phonautograms into audio, making him retroactively the earliest recorded human voice.

Where the credit landed

Scott recorded sound before anyone and reproduced none of it, which makes him a genuinely hard case rather than a wronged one. His own framing, that he had invented sound recording and Edison merely added playback: understates how large that addition was. The interesting point is that the record existed, unheard, for a century and a half, waiting for a technology able to read it.

Known For

Recording and playback are separate inventions. This is the cleanest proof.
The 1860 phonautogram was first heard in 2008.
He objected to Edison receiving the credit, and was partly right.

Knowledge Base

What You'll Learn

A recording nobody could play is still a recording.

How a 19th-century soot trace became audio

The First Sounds researchers photographed the phonautograms at high resolution and had software follow the traced line, converting its vertical displacement over time into a waveform. Nothing physical touched the fragile paper. The same optical approach is now used to recover audio from damaged wax cylinders and broken discs that could never survive a stylus.

Where This Fits

The phonautograph wrote sound down without playing it back, an early capture link in the chain followed in A History of Broadcast Technology.

FAQ

Q: What is the oldest recording of a human voice?
A: A phonautogram made by Scott de Martinville in 1860, of a voice singing "Au clair de la lune". It was never played in his lifetime. Researchers recovered the audio by optical scanning in 2008.

Q: Why did he not play his recordings back?
A: He was not trying to. The phonautograph was intended to produce a visual record of sound for study, in the way a seismograph records tremors. Reversing the process to recreate the sound was Edison's contribution.

Sources

The facsimile of Scott's own manuscript was opened and its title page read. Checked 20 August 2026.

Édouard-Léon Scott de Martinville, Principes de Phonautographie, 1857. The title page states that the original is preserved at the Académie des sciences de l'Institut de France as number 1639 of 1857. Digital images were made at the Academy in March 2008 and the facsimile edition was published by David Giovannoni for First Sounds in April 2010, under a Creative Commons licence.
Édouard-Léon Scott de Martinville, Fixation et Transcription du Chant, 1860. Described in the First Sounds catalogue as an album of phonautograms made in March and April 1860, which is the anchor year above and the source of the recovered recording. The file was too large to open here, so this description comes from the publisher's own index rather than from the document.