Sound is the name we give to certain frequencies of fluctuating pressure waves which resonate the human eardrum and are able to be perceived by the human brain. Waves fitting this description have existed since the big bang, all of them dissipating forever into space without any decipherable trace.
Towards the end of his life radio pioneer and otherwise shrewd businessman Guglielmo Marconi imagined that he would be able to recover these sounds, given a powerful enough receiver and a device able to filter them. It’s an appealing notion, that we can hear Bach or Julius Caesar or any of the musicians working before music began being written down using full notation (only 500 years ago), but of course such a thing is impossible. As soon as waves are created, they begin to dissipate, intermingle, bounce around and echo – these resonances aren’t just a vital element of every sound we hear; in many ways, they are the sound we hear.
The story of sound, told in sound, has to begin with its earliest capturing. Now, the story you may have heard about the birth of sound recording goes something like this; Thomas Edison, alone in the lab after a hard day’s work in 1877, manages to record a recital of “Mary Had A Little Lamb” onto a wax cylinder. You may have even think you have heard the recording, but I can guarantee you haven’t. The recording in circulation comes from a 1927 recreation for the Golden Jubilee of the Phonograph. The original was lost on a sheet of re-usable tinfoil fifty years earlier.
But it really doesn’t matter. The real start date for us is nearly a quarter of a century earlier, in the studio of French printer and bookseller Édouard-Léon Scott de Martinville. The year was 1853 or 1854, and he was working on engravings for a physiology textbook, in particular a diagram of the internal workings of the human ear. What if, he thought, we could photograph sounds in the way we do images? (photography was a quarter-century old at this point) He began to sketch a device, a way of mimicking the inner workings of the human ear in order to make lines on a piece of paper.
I cover a plate of glass with an exceedingly thin stratum of lampblack. Above I fix an acoustic trumpet with a membrane the diameter of a five franc coin at its small end—the physiological tympanum (eardrum). At its center I affix a stylus—a boar’s bristle a centimeter or more in length, fine but suitably rigid. I carefully adjust the trumpet so the stylus barely grazes the lampblack. Then, as the glass plate slides horizontally in a well formed groove at a speed of one meter per second, one speaks in the vicinity of the trumpet’s opening, causing the membranes to vibrate and the stylus to trace figures.
The result of this wasn’t a reproducible waveform, of course, such a thing was neither imagined, nor intended. As with Edison later, the primary practical use of this new science was for stenography. Édouard-Léon Scott de Martinville imagined that we could take these sounds and use them to identify single letters, or phonemes (something which is still pretty tricky now) – here is an example of an attempt to match word to sound trace – a truly futile project at the time.
The phonoautogram, as he called it, went through a number of prototypes, with the horn being replaced with a barrel and the sheet of glass replaced by a roll of parchment coated in lampblack (soot) in which markings could be made with a single thread from a feather. It wasn’t until 1859, however, when with the help of Rudolph Koenig he built a reliable version of the machine, and transcribed the sound of a tuning fork, that the recorded sounds become truly decipherable.
By “decipherable” I mean this – sitting in a vault in Paris for 150 years before the amazing people at Firstsounds.org (David Giovannoni, Patrick Feaster, Richard Martin and Meagan Hennessey) decided that recovering something from these scraps of paper was a project worth pursuing. Re-engineering these tracings into sound waves was a painstaking, technical business, but finally a recording was ready for the world – the sound of (probably) Scott himself singing the French folk song Au Clair de la Lune, recorded on the 9th of April 1860.
The release of this recording, stretching our record of sound back nearly two decades, caused a minor sensation. Here is Charlotte Green on BBC Radio 4 reporting the news – the giggling fit afterwards is due to another guest in the studio saying it sounded like a bee was trapped in their headphones.
The best telling of the story of how we came to have these recordings is from 2012, when the PRI radio show Studio 360 ran a feature on the recovery of the audio. This extended version of a fasinating, comprehensive sub-10-minute primer on the subject.
So here it is, our first mix. This is the second version of this mix – the original contained only the 1860 recordings, but since then a number of others have emerged online, including a restored recording of a cornet from 1857.
Tracklisting and details
1. Diapason at 435 Hz–at sequential stages of restoration (1859 Phonautogram) 0:00
This is an example of the tuning fork tone which made the restoration possible. We start from noise, and from the noise, a sound emerges.
2. Notes played on guitar by Adolphe Giacomelli (1853 or 1854) 0:24
3. First ever voice recording captured from the air (1853 or 1854) 0:29
These are some of Scott’s earliest experiments, made on his earliest prototype machine.
4. Phonautography of the voice at a distance (March 1857) 0:30
5. Song of the voice, changes in tone (July 1857) 0:34
6. Song at a Distance (“The Echoes”) (August 17, 1857) 0:51
7. Ashen Pipe (Aug – Oct. 1857) 1:07
8. Stylus of Bristle (Aug – Oct. 1857) 1:35
9. The Sound of a Deep Voice (October 1857) 1:46
10. The Lord’s Prayer (October 1857) 1:58
11. Study of the Timber of the Voice (November 1857) 2:20
These are recordings made on the earlier version of the phonoautogram machine, mostly indecipherable as the speed was not properly regulated. The “Song at a Distance” possibly features the voice of a young girl, and should it later be decoded it may count as the first recorded music.
12. The Timber of the Cornet (December 1857) 2:31
This is now the earliest fully-recovered recording. The soloist is unknown, but the tone of the instrument is unmistakable.
13. Au Clair de la Lune (April 9, 1860) 2:45
14. Shakespeare : Othello excerpt (April 17 1860) 3:13
15. R, I, RI, R, A, RA, RIRA (Will Laugh) (April 18, 1860) 3:21
16. Racine : Phedre (excerpt) (April 19, 1860) 3:27
17. Tasso : Aminta (excerpt) (April – May 1860) 3:44
18. Vocal Scale (May 17, 1860) 3:57
19. Cherubini : Et Incarnatus Est (Sept 1, 1860) 4:12
20. Masse : Fly, Little Bee (September 1860, or later) 4:25
These are the fully-restored recordings from the final version of the phonoautogram. Note the difference in the sound of “Au Clair de la Lune” compared to the radio broadcast. At the time it was believed to be the recording of a young girl – now it is thought to be Scott’s voice.