THE ARTICLE · 8 MIN
Ask who first recorded sound and the usual answer is Thomas Edison. His own national historical park says otherwise: “That honor goes to Edouard-Léon Scott de Martinville, a French inventor who in 1857 devised his phonautograph”. The difference is playback. Scott could record sound, but not hear it again; Edison built the first machine that did both. Here is how recording began, from the patents, the archives and the people who recovered the earliest recordings.
A short timeline
| Year | What happened |
|---|---|
| 1857 | Scott patents the phonautograph, which traces sound on paper |
| 1860 | Scott records “Au clair de la lune” |
| 1877 | Charles Cros deposits a sealed note describing playback (April); Edison builds his tinfoil phonograph (December) |
| 1887 | Emile Berliner patents the “Gramophone” |
| 1925 | Electrical recording goes into regular use at major US record companies |
| 1935 | The Magnetophon tape recorder is shown in Berlin |
| 1971 | The first commercial digital recording is released in Japan |
| 2008 | Scott’s 1860 recording is played back for the first time |
1. The phonautograph: recording without playback (1857)
Scott’s patent papers were deposited in Paris on 24 March 1857. In his own later note, he refers to “mon brevet d’invention délivré le 25 mars 1857 sous le nº 31470” — his patent issued on 25 March 1857, number 31,470.
His machine turned sound into a visible trace. It was meant to be looked at, not listened to. First Sounds, the group that recovered his recordings, is plain about this: “Scott did not intend his phonautograms to be played back”. The Edison park agrees, calling it “a record-only device”.
It still had a long life. First Sounds notes: “As a laboratory instrument it contributed for decades to the nascent science of acoustics.”
2. “Au clair de la lune”, 1860 — and the playback that got the speed wrong
“Scott recorded the French folksong “Au Clair de la Lune” on April 9, 1860”. According to First Sounds, “It remains the earliest clearly recognizable record of the human voice yet recovered.” Older phonautograms survive too, some from as early as 1853–54. First Sounds adds: “Many phonautograms from 1857 also survive, but they lack the tuning-fork timecode” needed to play them at the right speed.
In 2008, First Sounds released sound files made from Scott’s paper traces using “the Lawrence Berkeley National Laboratory’s virtual stylus technology”. The first release sounded like a woman or a child singing. The speed turned out to be wrong. The team later explained: “we chose a playback speed that presumed the singer was a woman or an adolescent”, then “we realized that this speed was too fast”.
The clue was a recording of speech. Scott had written that his tuning fork made “500 simple vibrations per second”. In early 2009 the team played a spoken recording at the speed used for the 2008 release, and it “sounded very much like a tape played back at twice the proper speed”. Instead, “adjusting the timecode to 250 Hz gave us a natural-sounding low male voice”. Patrick Feaster of First Sounds explains that, “despite Scott’s own statements to the contrary”, the phrase “would clearly have meant 250 Hz in the terminology of mid-nineteenth century acoustics”. Comparing it with another of Scott’s recordings, the researchers concluded that the singer sounds “very much like the speaker” in a recording Scott made of himself. So it is probably Scott singing, though that is a voice-match judgement rather than proof.
Even the words were first misheard: the recording was later found to be “Au clair de la lune, mon ami Pierrot”, “rather than “Au clair de la lune, Pierrot répondit,” as originally announced”.
3. The playback idea, on paper first: Charles Cros (1877)
The idea of playing back a recording was written down in France before Edison built his machine. The French inventor Charles Cros sent a sealed note to the French Academy of Sciences, “deposited at the French Academy of Sciences on April 30, 1877”. It was opened and read on 3 December 1877. In it he described recording the movement of a vibrating membrane and using that trace “pour reproduire le même va-et-vient” — to reproduce the same movement.
But the note was sealed, so Edison could not have seen it: “Cros’ proposal was in a sealed envelope held in confidence by the French Academy of Sciences.” And “Cros never built a paléophone, nor was a paléophone ever built to his specifications.” The name paléophone came later: “for most of 1877 Cros had no name for his idea.”
4. Edison’s phonograph (1877)
Edison’s idea came out of telephone work. According to the Thomas A. Edison Papers at Rutgers, “In July 1877, while developing his telephone transmitter, Edison conceived the idea of recording and playing back telephone messages.” His machinist John Kruesi built the tinfoil phonograph “during the first week of December 1877”.
Edison’s national park adds a detail the usual story leaves out: “this iconic machine was not the first phonograph – but it was the first to be shown outside the lab”. An earlier paper-strip device from July 1877 “remained just a prototype in the lab”. On 7 December 1877 Edison and two colleagues took the tinfoil machine to Scientific American, whose editors reported that “the machine inquired as to our health, asked how we liked the phonograph”.
His patent application was “filed December 24, 1877”, and the patent, No. 200,521, is “dated February 19, 1878”. Its purpose: “to record in permanent characters the human voice and other sounds”.
What about “Mary had a little lamb”? The story is real but more tangled than “the first words ever recorded”. Rutgers records Edison’s recollection that he “recorded the nursery rhyme “Mary Had a Little Lamb"" on the tinfoil machine. His colleague Charles Batchelor, writing decades later, remembered Edison reciting it, “one of our favorite stereotyped sentences used in experimenting on the telephone”, into the earlier paper-strip prototype, which played back “ary ad elll am”. And according to the national park, Kruesi’s machine went first to Batchelor, “who tested and adjusted the finicky mechanism before showing it to Edison”. Both accounts are later recollections.
Did Edison know about Scott? The park says: “Primary sources are silent on whether Edison even knew of the phonautograph in 1877.”
5. Discs: proposed in 1877, made to win by Berliner
Emile Berliner’s gramophone is remembered as the start of disc records. The history is not that simple. Edison’s park notes that “In 1877 and 1878 plate (later called “disc”) designs appeared in laboratory sketches and patent applications nearly as much as cylinder designs”, including a British patent of April 1878.
Berliner’s first gramophone patent, “GRAMOPHONE. No. 372,786. Patented Nov. 8, 1887.”, describes wrapping the record “tightly about the drum”, and says it works “all substantially in the manner of the well-known phonautograph by Leon Scott”. A later patent, dated 15 May 1888, describes a surface “which may be a flat disk or a cylinder”. The disc format won in the end: its “disc records became synonymous with sound playback” for most of the 20th century.
6. Electrical recording (1925)
Before this, records were made by the acoustic method. Western Electric’s electrical process changed that. It was offered to Victor and Columbia “in the fall of 1924”, and “Columbia promptly signed with Western Electric, followed a few weeks later by Victor.” “Electrical recording at Columbia began on a regular basis in April” 1925 — though “neither Columbia nor Victor billed their new releases as electrically recorded until 1926”.
In a 1926 paper, J. P. Maxfield and H. C. Harrison wrote that with the new system “a range of frequencies from 60 cycles to 6000 can be recorded with reasonable success”.
7. Magnetic recording and tape
Magnetic recording is older than tape. Valdemar Poulsen’s US patent for magnetic recording was filed in 1899 and published in 1900. Practical coated tape came from Germany. (An American, Joseph O’Neill, applied in 1926 for a US patent on a paper strip carrying “a trail or line of magnetic material”, granted in 1927, but “did not pursue it commercially”.) Fritz Pfleumer patented a “sound paper machine”, filed in 1928, and licensed it to AEG. BASF says “The first 50,000 meters of magnetic audio tape are supplied to the electronics corporation AEG in 1934”, and according to the Computer History Museum, “The Magnetophon K1 recorder and Type C tape debuted at the Berlin Radio Show in August 1935.”
8. The first digital recordings: Japan first
Commercial digital recording did not start in the US. Denon (Nippon Columbia) worked with a recorder from the Japanese broadcaster NHK: “In 1969-1971, Denon leased an NHK stereo PCM recorder and conducted numerous test recordings.” According to a history published in the ARSC Journal, “the first commercial digital recording was Nippon Columbia NCB-7003, “Something” by Steve Marcus, released January 1971”.
In the US, a 1976 digital test recording at the Santa Fe Opera was not released from the digital tape: the album “was made from a mix-down of Bruck’s 16-track analog tape.” Telarc’s April 1978 recording with the Cleveland Symphonic Winds became “the first classical intended-for-release commercial digital recording in the U.S.”
Our reading
“Who invented sound recording?” has at least three fair answers: Scott first recorded sound, Cros first described playing it back, and Edison first built a machine that did both. Most “firsts” in this story survive only with a qualifier attached — first recognisable voice, first commercial digital album, first US classical one. Those qualifiers are where much of the history is.
Sources
- Thomas Edison National Historical Park (US National Park Service): “Origins of Sound Recording” — the inventors; Scott de Martinville; Charles Cros; Edison’s path to the phonograph; Thomas Edison; frequently asked questions.
- First Sounds: Scott’s phonautograms; “Earlier playback” (speed reinterpretation); Patrick Feaster, “Édouard-Léon Scott de Martinville: An Annotated Discography”, ARSC Journal 41:1 (2010).
- Thomas A. Edison Papers, Rutgers University, “Tinfoil Phonograph”.
- US Patent 200,521 (Edison, 1878); US Patents 372,786 (1887) and 382,790 (1888) (Berliner); US Patent 661,619 (Poulsen, 1900).
- Comptes rendus hebdomadaires des séances de l’Académie des sciences, t. 85, séance du 3 décembre 1877 (Cros).
- Discography of American Historical Recordings, UC Santa Barbara, “Columbia Corporate History: Electrical Recording and the Late 1920s”.
- J. P. Maxfield and H. C. Harrison, “Methods of High Quality Recording and Reproducing of Music and Speech Based on Telephone Research”, Bell System Technical Journal 5:3 (1926).
- Computer History Museum, “1935: Audio recorder uses low-cost magnetic tape”; BASF, corporate history, 1934.
- Thomas Fine, “The Dawn of Commercial Digital Recording”, ARSC Journal 39:1 (2008).
Checked September 2026.
Related: Music myths checked · Older than you think
- music
- history
- inventions
- sound recording