The oldest recorded human voice that can be heard intelligibly today dates from April 9, 1860: a ghostly snippet of the folksong “Au Clair de la Lune,” played back in March 2008 after lying silent for nearly a century and a half. For twenty seconds, someone living in the Second French Empire of Napoleon III had sung into a funnel while a stylus attached to a membrane at its far end was tracing a line on a paper blackened with soot and wrapped around a rotating cylinder. The resulting squiggle represented the sound waves of the singer’s voice as a graph of amplitude against time, much like the waveforms displayed on computer screens by modern sound-editing programs. Years later, the First Sounds initiative found that squiggle in the archives of the French Academy of Sciences and succeeded in converting it digitally back into intelligible sound.[1]
But this was not at all the outcome Édouard-Léon Scott de Martinville had in mind back in 1860 when he recorded “Au Clair de la Lune” on his recently invented phonautograph. His goal was instead to capture the voice automatically in a way that could be read and appreciated by the eye. Critics today who take the reproduction of recorded sounds for granted often have a hard time grasping Scott’s motives; for instance, one of them has playfully characterized the phonautograph as “designed to record sounds but not to play them back, making it both the world’s first and most useless recording device.”[2] But for a world that had no experience of playback to guide its expectations, Scott’s invention was no less complete in its functionality than a seismograph that records earthquakes but cannot recreate them. From the perspective of 1860, the feat of making the evanescent nuances of the human voice fix themselves on a sheet of paper was no weak half-measure, but a triumph of ingenuity with stunning implications.
In American tradition, for instance, the idea of representing the human voice visually was then still a wild fantasy associated with the cycle of comic legends surrounding Davy Crockett. The Crockett Almanacs published between 1835 and 1856 not only fostered and shaped the image of Crockett as an American frontier hero but also introduced another character, Ben Harding (often “Hardin”), a sailor, as Crockett’s sidekick. The entry in which Crockett recounts his first meeting with Harding—in an almanac for 1839—ends with the following statement:
He said his name was Ben Harding, and then he telled such stories about what he had seed as made the gals dream o’ nights for a fortnite arter he was gone; and as I spose the reader would like to hear some of ’em, I think I shall put ’em in print; but he had a voice that was so ruff, I can’t rite it doun, but I will have a cut made to picktur it out. Here it is:[3]
The accompanying woodcut, captioned “Picture of Ben Harding’s Voice,” shows the character’s face in profile, his open mouth spewing forth what looks like a spray of shards of broken glass. The impact of this woodcut on Almanac readers varied. On one hand, it supported a claim that Harding’s voice exceeded the capacity of ordinary written description, adding to the many superhuman attributes assigned to the man himself as a heroic frontiersman. Thus, a British traveler visiting the United States in 1839 recounted hearing that Crockett’s own voice, rather than Harding’s, had been “so rough it could not be described—it was obliged to be drawn as a picture.”[4] In 1858, however, a Gettysburg newspaper alluded to the woodcut from another perspective: “Some of our readers have seen in old ‘Crockett Almanacs’ a picture of ‘Ben Hardin’s Voice,’ a storm of corners, points and rough edges, and no doubt laughed at the oddity of the idea of making the tones of the voice visible.”[5] In this case, Crockett’s graphic representation of the roughness of Harding’s voice was considered as remarkable as the voice itself. The woodcut of 1839 had embodied not one but two comically implausible claims: that a voice could be so rough as to require being “drawn as a picture” and that a voice could be pictured in this way.
But could it, in fact, given the inexorable march of technological progress? The introduction of photography as an automatic means for recording the visible world inspired critics to imagine analogous technologies for recording other subjects, including human speech. Some writers in the first half of the nineteenth century identified phonetic shorthand writing as “the daguerreotype of sound,”[6] or as “the daguerreotype of literature” which “spreads thoughts on paper as rapidly as the daguerreotype depicts forms on iodine,”[7] but others imagined some new instrumentality altogether: a speech-recorder analogous to the camera, but technologically distinct from it. Commodore Matthew F. Maury, superintendent of the Naval Observatory in New York, made one such suggestion in a letter of May 12, 1844:
What a pity it is that M. DaGuerre, instead of photography, had not invented a process of writing by merely speaking through a trumpet upon a sheet of paper. What a glorious thing it would have been. I could then have mailed out letters in the boldest hand and at any time. Instead of saying “I wrote you a letter last Monday,” the phrase would have been “I spoke you a ream last Tuesday.” The world would become a mere scribbling shop—a vast book machine.[8]
A short story by J. H. Ingraham, published two years earlier, describes a fictional instrument known as the acoustype: words spoken into a spiral platinum tube passed through a floss silk “lens” which focused them in such a way that, on striking a sheet of paper, they left legible inscriptions in pink.
Do you not see that it does away with the writing master’s profession—renders writing wholly unnecessary. With a portable acoustype in your hand and a sheet of letter-paper before you, you have only to speak your thoughts through it and you imprint them upon the paper as fast as you can speak. Think, sir, what an immense value it will be to authors, to public speakers! Yes, sir, a member of Congress can make his speech and put it upon paper all in the same breath, with a little boy before him to turn over the leaves.[9]
Maury and Ingraham were hoping merely for a labor-saving device that would automatically convert speech into conventional script—which, for them, would have been quite impressive enough. In 1847, however, the poet and journalist Théophile Gautier responded to the death of the famous actress Mademoiselle Mars by speculating: “Just as light has been forced to blacken a plate with images, so will it be found possible to keep sound waves and thus to preserve the execution of an air of Mario, a tirade of Rachel, a couplet of Frédérick Lemaître.”[10] In a similar flight of fancy, Ralph Waldo Emerson stated in 1851: “The sun paints; presently we shall organize the echo, as now we do the light shadow.”[11] The future could then enjoy portraits of people’s voices along with their photographs; the audible part of their genius and personality would no longer die with them.
So these were some of the ideas in circulation when the Parisian typesetter Édouard-Léon Scott de Martinville had his flash of inspiration:
In 1852, correcting the proofs one day at the Martinet printing house for the first edition of the Treatise on Physiology by Professor Longet, the idea came to him of applying the acoustic means realized by nature in the human ear to the graphic fixation of the sounds of the voice, song, and instruments. He calculated to arrive in this way at an acoustic stenography of speech, without the assistance of the hand of man.[12]
He made his first artificial ear experiments in late 1853 or early 1854, trying to record the human voice and the sound of a guitar on lampblacked plates using styli attached to paper membranes analogous to the human eardrum. On January 26, 1857, he deposited those experiments with the French Academy of Sciences, attached to the end of an essay on what he called the “principles of phonautography,” literally the self-writing of sounds. He begins this essay with the familiar analogy of acoustic photography and then proposes several use scenarios for it:
Is there a possibility of reaching in the case of sound a result analogous to that attained at present for light by photographic processes? Can one hope that the day is near when the musical phrase, escaped from the singer’s lips, will be written by itself and as if without the musician’s knowledge on a docile paper and leave an imperishable trace of those fugitive melodies which the memory no longer finds when it seeks them? Will one be able to have placed between two men brought together in a silent room an automatic stenographer that preserves the discussion in its minutest details while adapting to the speed of the conversation? Will one be able to preserve for the future generation some features of the diction of one of those eminent actors, those grand artists who die without leaving behind them the faintest trace of their genius? Will the improvisation of the writer, when it emerges in the middle of the night, be recoverable the next day with its freedom, this complete independence from the pen, an instrument so slow to represent a thought always cooled in its struggle with written expression?[13]
Scott’s “fugitive melodies” may have been either improvisations in vocal music or virtuosic interpretations of written scores, but his proposed preservation of the “diction” and “genius” of actors was clearly focused not on mere words but on the paralinguistic and prosodic aspects of their enactment in live performance. In March 1857, Scott formally patented a model of phonautograph that made its inscriptions on flat glass plates and relied on a pair of membranes separated by an airtight space, analogous to the eardrum and oval window of the middle ear.
“Phonautography of the human voice at a distance,” included with Scott’s 1857 patent papers
Louis Figuier was one of the first writers to comment publicly on Scott’s phonautograph, and some of his expectations closely mirrored those of the inventor. He praised the instrument as a valuable tool for scientific experimentation and a potential automatic stenographer, but like Scott he also suggested that phonautograms would capture other previously undocumentable nuances of the living voice:
Writing and print express speech, it is true, but speech dead and faded. You chance to hear some beautiful verses recited by Rachel: write them down and give them to a child to read, you will no longer recognize them. To return the life to them, it would have been necessary to accentuate them, to note them as in music; still, the goal would only have been very imperfectly reached. Something is missing there; it is what many enlightened men feel, but without hope of filling the gap. The phonautography of M. Scott will provide the means of printing with ordinary script the expression which it lacks, that is to say, of representing thought graphically by the expression of the word; because the amplitude of the graphic trace, or the weak measurement of this same trace, would correspond exactly to the various inflections of the voice by which the declamation is accompanied.[14]
True, he concedes, there had already been efforts to devise manual systems for noting down the rhythm and pitch of speech “as in music”; Joshua Steele had set one forth in 1775, which James Boswell wished had been used to preserve the “mode of speaking” of Samuel Johnson for posterity.[15] But Figuier asserts that this approach still left out something important. His reference to “Rachel” must have struck readers as particularly poignant: Élisabeth Rachel Félix was a celebrated actress who died on January 3, 1858, after a long struggle with consumption. “Of the burning intensity which characterized her rendering of passion in its fiercer concentrations, no words can give an adequate image,” it was reported elsewhere.[16] Indeed, a decade earlier, Théophile Gautier had listed “a tirade of Rachel” among the subjects to which he would have liked to apply a photography of sound. If only she had made phonautograms, Figuier suggests, something of her genius could have remained alive.
The idea that phonautograms would preserve subtle nuances of speech, and not just words, soon spread to English-language publications. An American newspaper, recalling the woodcut of Ben Harding’s voice published in the Crockett Almanac, announced that Scott had now “made the absurdity of the almanac’s idea a reality” by producing “a sort of picture of the voice.”[17] A London Review article observed further that the aesthetic impact of a phonautogram appeared to match the character of the sound that had produced it:
Every separate source of sound has an individuality of its own. The sounds of different musical instruments, for instance, are easily distinguished from one another, and from the human voice. This latter, moreover, gives different traces, according to its character,—the sweet, soft voice of a female, especially when singing, being characterized by great beauty and harmony in the curves impressed on the paper; in those produced by the harsher voices of a man, the curves are larger and more rugged looking; whilst in a shriek or a shout, or in the harsh, discordant sounds of instruments, the waves are irregular, unequal, and broken up into secondary vibrations of all degrees of amplitude.
Phonautograms were reassuringly consistent with gender stereotypes: traces of the female voice showed “great beauty and harmony,” while traces of the male voice were “larger and more rugged looking,” approaching the Crockett Almanac woodcut of Ben Harding’s rough voice. The voices of people in unsettled states of mind produced traces that were chaotically “broken up.” Different styles of oratory also manifested themselves in visually appropriate ways, translating peculiarities of utterance into conventions of text:
An oration, delivered with varying rapidity, and with the pitch of the voice greatly modulated in different parts, has a very striking appearance in its phonograph. Rapidly spoken parts have the curves crowded together, whilst in others they are widely separated. The loud tones of the voice are shown by the written waves rising to perhaps half an inch, or more, in height, whilst the low tones are not more than the eighth of an inch high; the modulations of the voice are thus shown very beautifully by the varying height of what may be called the letters of sound.
Loud speech was thus represented naturally by large “letters”; rapid speech was suitably “crowded.” Unfortunately, the very individuality that made these recordings such perfect indices of their sources seemed to make it difficult to translate them into legible words, although not necessarily impossible:
Not only does the impression vary with the tone of the voice, the rapidity or loudness of utterance, but it has been found that the same words uttered by one person are written down by the instrument very differently from the way they are when spoken by another; just as the handwriting of one person differs from another. This, however, is a difficulty which will be overcome by practice, and perhaps improved instrumental arrangements: even now we hear that Mr. Scott is attaining some facility in reading off these natural stenographs.
The phonautogram was, paradoxically, a kind of inscription that was both more and less transparent than ordinary print. It bore the stamp of its speaker’s gender, mental state, and style, but the words themselves were extraordinarily challenging to identify. If the art of deciphering the words of phonautograms were developed further, however, the writer thought Scott’s system would revolutionize the practice of turning speech into text, “whether it be in respect to the unimpeachable accuracy of the process; the entire absence of trouble and expense in reporting any articulate sounds; or the great saving of the time and the exhausting labors of our parliamentary reporters.”[18]
Scott obtained a certificate of addition on his original patent on July 29, 1859, describing a new design with a single membrane and a paper sheet wrapped around a cylindrical drum, the form in which his invention is best known. He continued to claim that his invention could be used to transcribe the nuances of great acting and oratory, but he was now careful to differentiate this goal from the less promising one of distinguishing words. In the addition to his patent, he included a page headed “Application of phonautographic notation to the transcription of elocution” in which he sets forth his revised aspirations:
For noting elocution exactly, it is not enough to mark at the top or the bottom of the line the longs and shorts, the fortés and the pianos, the rises and falls of the tone, the inhalations, the breathing, and the pauses and the outbursts. It is necessary to represent clearly and easily the quantum or the mathematical value of each of these modifications.
The phonautographic trace furnishes at present—without one having to be occupied with articulation [i.e., the ability to distinguish words]—a very simple means of objectively representing the artist’s diction. This trace is a kind of reptile, the coils of which follow all the modulations or inflections of discourse. It suffices for translating by sight—except for the articulation—to make the following remarks: the horizontal distance of the foot of the curves indicates the pitch or tonality; the height of the same curves the intensity of the voice; the detail of the curves the timbre; the absence of curves the pauses or silences. The few natural expressions opposite suffice for understanding this page.
To the right Scott drew some sample waveforms with captions identifying the kind of speech to which each corresponded: “Represents the deep voice,” “the high-pitched voice,” “a high-pitched voice descending to a deep one,” “a deep voice rising to the high-pitched one,” “an intense voice,” “an average voice,” “a weak voice,” “the trill on the letter r,” “the cadence on a vowel,” and “the outburst of the voice.” Finally, Scott devoted the bottom of the page to a sample interlinear text comprising seven lines excerpted from Jean-François Ducis’s Othello. To the left, in each case, was a line in conventional script. In the center of the page was, first, a condensed phonautogram of the artistic oral delivery of the same line. Underneath, Scott then wrote out the line in a phonetically simplified version of French, varying the height of the letters to match the amplitude of the phonautogram and otherwise encoding supposed phonautographic characteristics into the script, such as wiggliness to represent a rolled r, producing a sort of hybrid inscription.[19] The reader of such a text could arguably have experienced something akin to “hearing” the original performance.
But in the end, nobody ever learned to read phonautograms as Scott had originally hoped. Expert acousticians found misguided reasons to argue that the phonautograph could not record complex sound waves from the air after all.[20] Rudolph Kœnig, who obtained an exclusive license to manufacture the phonautograph for sale, pursued it only as a dry scientific instrument for the study of tuning forks, organ pipes, and the like—not as the medium of voice and performance of which Scott had dreamed. In 1861, Scott made one final effort to put his phonautographic work on record before abandoning it for a career as a bibliographer. He gathered together a sampling of his latest and finest inscriptions and deposited them with the French Academy of Sciences: vocal scales, excerpts of songs, a dramatic recitation in Italian.[21] He hoped posterity might one day see these inscriptions; we have now heard one of them, which is admittedly a big difference. And yet the playback of “Au Clair de la Lune” also fulfills the inventor’s dream in a sense that is perhaps more important than the distinction between reading by eye and listening by ear. Scott believed he was capturing subtleties of the living voice that were missing from conventional writing and musical notation—the same subtleties, in principle, that can still make a sound recording more valuable than a score or a script today. Now that we’ve heard the ghostly song from 1860, we know he succeeded.
Notes
This essay, posted here to commemorate the 200th anniversary of the birth of Édouard-Léon Scott de Martinville (April 25, 2017), was originally published as Patrick Feaster, “Daguerreotyping the Voice: Léon Scott’s Phonautographic Aspirations,” in Parole #1: The Body of the Voice / Stimmkörper, edited by Annette Stahmer (Cologne, Germany: Salon Verlag, 2009): 18-23. However, please be aware of the following updates and corrections:
- “The oldest recorded human voice audible today” → changed to “The oldest recorded human voice that can be heard intelligibly today”; even older voice phonautograms recorded by Scott have surfaced and been educed in the meantime, but because they lack an accompanying tuning-fork trace as a guide to speed correction, they remain distorted beyond recognition by severe speed fluctuations.
- “laying silent” → corrected to “lying silent” (grammar)
- “ten seconds” → corrected to “twenty seconds”; the initial playback current at the original time of writing was later found to have been set to twice the correct speed. The audio is presented in this blog post at the correct speed.
- “someone living in the Second French Empire of Napoleon III”; now identified with reasonable certainty as the inventor himself, recording his own voice—a point formerly obscured by the earlier, incorrect playback speed—but since he was still “someone,” I’ve left the original text unaltered.
- “just like the waveforms” → “much like the waveforms”; there is a technical difference, outlined here, although the audio as presented in this blog post actually treats them identically.
- “Maury and Ingram” → “Maury and Ingraham“; the author’s name had been misspelled in one of my sources, and while I’d fixed one appearance of it, I’d missed another.
- “The work in question is presumably” → “The work in question is,” confirmed beyond reasonable doubt.
- Citations to retired First Sounds Working Papers have been replaced by citations to our currently maintained translation and facsimile documents.
- I’ve since tracked down the original sources for some subsequently reprinted quotations (e.g., Matthew Fontaine Maury to Ann Maury, May 12, 1844, published in Diana Fontaine Maury Corbin, A Life of Matthew Fontaine Maury (London: Sampson Low, Marston, Searle, & Rivington, 1888), 47), but I’ve retained my earlier citations below.
[1] For the recording of “Au Clair de la Lune” and more information about this initiative, see FirstSounds.org.
[2] Adam Sternbergh, “In Your Face, Edison: New Oldest Recording Found,” New York Magazine, March 27, 2008, http://nymag.com/daily/entertainment/2008/03/edison_pwned.html, accessed May 6, 2008.
[3] Michael A. Lofaro, ed., The Tall Tales of Davy Crockett: The Second Nashville Series of Crockett Almanacs, 1839-1841 (Knoxville: University of Tennessee Press, 1987), 1839:24.
[4] R. G. A. Levinge, Echoes from the Backwoods, or Sketches of Transatlantic Life (London, 1849), 2:11-13, quoted in Franklin J. Meine, ed., The Crockett Almanacks: Nashville Series, 1835-1838 (Chicago: Caxton Club, 1955), v. Meine concludes from internal evidence that Levinge’s travels took place in 1839.
[5] “Daguerreotyping the Voice,” Compiler (Gettysburg, Pennsylvania), June 21, 1858, p. 1.
[6] “Letters from New York,” Southern Literary Messenger 15 (Mar. 1849), 187.
[7] Extract of a speech “delivered by Mr. J. [sic] Pitman, at a Phonographic Soiree, held in the Council Chamber of the Town Hall, Ipswich, England, May 14, 1845,” printed as “Phonography,” Liberator 15 (July 18, 1845), 116.
[8] Quoted in “A Curious Prophecy,” Phonogram 1 (Sept. 1891), 196.
[9] J[oseph] H[olt] I[ngraham], “Daguerreotype Sketches,” Ladies’ Companion 18 (Nov. 1842), 45-6.
[10] Quoted in Alfred Chapuis and Edmond Droz, Les automates, figures artificielles d’hômmes et d’animaux (Neuchâtel : Éditions du Griffon, 1949), 333, my translation.
[11] Ralph Waldo Emerson, “Address to the Citizens of Concord on the Fugitive Slave Law, 3 May 1851,” Houghton Library of Harvard University, bMS Am 1280.201 (22), 65r.
[12] Louis Figuier, “Causerie Scientifique,” Grande Revue 5 (Dec. 25 [?], 1890; reproduced in Thomas E. Jeffrey, ed. Thomas A. Edison Papers: A Selective Microfilm Edition [Frederick, Maryland: University Publications of America, 1985—], 146:636-44), 565, my translation. The work in question is F[rançois]-A[chille] Longet, Traité de physiologie , 2 vols. (Paris: Masson, 1850—).
[13] Patrick Feaster, Phonautographic Manuscripts, 5. As to the date of Scott’s first experiments, he cites “three” years ago here (4, 12), but “four” years ago in October (24, 39).
[14] The article is reprinted in Figuier, “Causerie,” 568-73, there identified as taken from the Année Scientifique de 1858 (3e année). The quoted portion appears on 572-3; my translation.
[15] Joshua Steele, An Essay Towards Establishing the Melody and Measure of Speech (London: W. Bowyer and J. Nichols, 1775); George Birkbeck Hill, ed., Boswell’s Life of Johnson (Oxford: Clarendon Press, 1971), 2:326-7.
[16] H. T. Peck, ed., The International Cyclopædia (New York: Dodd, Mead & Co., 1898), 12:366.
[17] “Daguerreotyping the Voice,” Compiler (Gettysburg, Pennsylvania), June 21, 1858, p. 1.
[18] “Sound Recording Itself,” from London Review, in The Living Age 70 (Sept. 21, 1861), 726-727.
[19] Patrick Feaster, Phonautographic Manuscripts, 66-67; David Giovannoni, ed., First Sounds Facsimile Series (FirstSounds.org: 2010—) 2:19.
[20] See e.g. Fr. Jos. Pisko, Die neueren Apparate der Akustik (Vienna: Carl Gerold’s Sohn, 1865), 75-78.
[21] Patrick Feaster, Phonautographic Manuscripts, 70ff. Scott did also return to the phonautograph at the very end of his life with a self-published volume entitled Le problème de la parole s’écrivant elle-même (Paris, 1878).
Griffonage-Dot-Com 
Pingback: All Griffonage That On Earth Doth Dwell | Griffonage-Dot-Com