School taught us that Thomas Edison invented the phonograph in 1877 and was the first person to create a sound recording that could be immediately played back. But school didn’t mention that people had been rendering music on paper for centuries, using graphic representations of frequency and duration that could be used to reproduce the sound.
These artifacts were essentially unplayable for centuries, but historian and ethnomusicologist Patrick Feaster has deciphered and recreated sound from these documents in Pictures of Sound: One Thousand Years of Educed Audio, a CD and book package released last year by the Dust-to-Digital record label. This set features computer-generated renditions of music dating back to 980 A.D.
Édouard-Léon Scott de Martinville, a French scientist who captured sound on paper in 1860.
The project began when Feaster teamed up with the label to produce a vinyl single of a fragment of “Au Clair de la Lune” that French scientist Édouard-Léon Scott de Martinville had captured on paper with a machine called a phonautograph in 1860, nearly two decades before Edison made his first audio recording.
Scott de Martinville wanted to preserve the performances of actors and singers with an audio equivalent of photography. He correctly deduced that he would need to create an “artificial ear” to capture the vibrations in the air that humans process as we listen to sound. His invention, the phonautograph, was the first machine that ever recorded sound waves over time, but he had only aimed to represent the sound on paper as a way of improving upon written language, which could not adequately convey tonality, intensity, or timbre. He had hoped that people would eventually learn to read phonautograms by eye, and translate the sound in their minds in a way similar to how we read words on a page.
Details of Scott de Martinville’s phonautogram of “Au Clair de la Lune.”
Scott de Martinville’s phonautograms were basically nothing but scientific curiosities until 2008, when David Giovannoni of the First Sounds initiative and the Lawrence Berkeley National Laboratory were able to convert digital images of them into digital sound files by using a “virtual stylus” to track the transcribed wave forms as if they were grooves on a vinyl record. “Au Clair de la Lune,” from April 9, 1860, was the first to be made into a crude audio file. Feaster then speed-corrected the track so it would be recognizable as a vocal rendition of the song.
Both of these recordings made by Feaster from Scott de Martinville’s phonautograms are presented three times in a row: the first time without speed correction, the second with speed correction, and the third time with speed correction, voice only, and in mono.
Automatic instruments and music boxes date back to the late 18th century. An early version of this technology known as tonotechny was a system in which a cylinder pricked with brass points or pins could be programmed to play the keys of an organ. These devices were not powerful enough to be useful in a large hall, but were used in parlors of homes. Programs for automatic instruments are not a form of sound recording, but the process of tonotechny aimed for a level of detail and precision, with the goal of emulating the performance style of a specific musician.
This sound recording was created from notage based on a song written and performed by Claude Balbastre, one of the most famous organists in Paris in the 1770s. The notation factors in subtle details of his playing to capture the essence of his performance style. This isn’t a recording of his playing, but it’s the next best thing.
An oscillogram by Leonhard Euler made in 1739.
Édouard-Léon Scott de Martinville was not the first to represent sound with an oscillogram — a graph of time versus amplitude. This recording created by Feaster recreates music from oscillograms transcribed by Claude-Servais-Matthias Pouillet in 1850, Thomas Young circa 1807, Leonhard Euler in 1739, and Francis North in 1677. Young’s illustration is the earliest known drawing of a waveform, which depicts time versus amplitude as a continuous undulating line.
Notation for a toothed cylinder.
The cylindrus dentatus — or toothed cylinder — was a 17th-century precursor to tonotechny. The cylinders were designed to rotate at constant speed while its teeth triggered the playing of notes on an organ. This recording educed by Feaster is based on notation crafted by 17th-century Jesuit scholar Athanasius Kircher for a song composed by Simplicio Todeschi that was published in 1627.
Feaster also attempted to educe sound from medieval notation that resembled graphs of time versus frequency more closely than modern staff notation. He educed this piece of music from medieval polyphonic notation found in a French manuscript dating to the middle of the 13th century.
In the liner notes for Pictures of Sound, Feaster admits that this is perhaps not a “meaningful representation of medieval music,” mainly because the notation doesn’t line up perfectly, which results in “very strange-sounding harmonies.”
Daseian notation, the oldest known written polyphonic music, was only used in a handful of works from the 9th century. One of the best known pieces presented in this notation is a treatise called “Scolica Enchiriadis.”
This eduction by Feaster may be the oldest automatically playable transcriptions of music in the world, and are seemingly accurate to the intention of the composer. The melody, harmony, and rhythm from Feaster’s eduction of the notation matches up modern choral performances based on the music.