It is easy to forget in this technological era that the latter half of the 19th Century was a major turning point between that of the old and the new. Politically, some defining trends of modernity, such as socialism, began to make their mark. Mans understanding of consciousness took a leap forward with the development of psychology, and physics also made its mark as this period began to usher in the very things that define the modern world for many people today.

Telephones developed rapidly, and motion pictures were shown commercially. Major cities began electrification and parallel developments in the electric motor led to unprecedented levels of industrialisation, and modern transport. It was into such a rapidly developing technological world that sound recording emerged, in a notably tentative fashion.

A late 19th Century street

 

Over 150 years ago audio recording became a workable reality. Until the last decade it was thought that Thomas Alva Edison was responsible for the very first recording, his recitation of Mary had a little lamb, dating to 1877. The recording no longer exists due to the fragility of the medium, the extant recording is actually a re-enactment of the event in 1927. When he announced the innovation later that year, it was so unexpected that he came to be known as “The Wizard of Menlo Park” due to the location of his research laboratory in New Jersey.

However, in 1859, a recording was made of a tuning fork, which still survives. A relatively accurate recording of the more complex acoustical phenomena of a human voice was made the following year. This seventeen/eighteen year time gap is a substantive one for such an era of rapid innovation. French inventor Édouard-Léon Scott de Martinville (1817 – 1879) recorded those very sounds onto a paper cylinder using a mechanical device called the phonautograph, which he invented. The phonautograph shouldn’t be confused with the phonograph, Edison’s subsequent invention.

Phonautograph, University of California

 

The Phonautograph explained

Essentially, the phonautograph functioned in a fashion that will be familiar to those that have read up on the way in which the microphone and the LP work. A thin membrane, akin to a sort of non-electric microphone, responded to the air pressure from nearby sounds. The membrane’s response in turn vibrated a stylus that marked sheets of paper, attached to a rotating cylinder. The sheets were blackened by oil-lamp smoke so the stylus would easily make a visible impression on them. These marks were a representation of the very sound waves that had first affected the membrane!

Scott de Martinville was familiar with a pre-existing physics theory that sound vibrations could be recorded as lines with a stylus when attached to a vibrating object. As early as 1807, Thomas Young described recording the vibrations of a tuning fork on the surface of a blackened cylinder. Rather than being motivated by an interest in abstracted physics, Scott de Martinville was focused on practical applications for sound recording, somewhat akin to that achieved with photography, a technology he was cognisant of since it had been developed in France, and yielded impressive results.

Scott de Martinville studied drawings of auditory anatomy and tried to mimic the workings of the ear mechanically. He substituted an elastic membrane for the eardrum. This is perhaps the most notable aspect of his invention for it was now possible to record sound vibrations from the air, without direct contact with the source of the sound, unlike the necessary connection of a tuning fork to Thomas Young’s prior device. This development is likely to have influenced early microphone invention in the later 19th Century, whereby a diaphragm responding to acoustic sounds is utilised to alter the magnetic field of a signal generator to create an equivalent electrical response.

Scott de Martinville’s method of sound recording was successfully patented in 1857. His first attempts to record date to late 1853 or early 1854, where he attempted to represent the sound of speech and guitar in graphical form. Inevitably however, the application of the phonautograph was very limited because very short segments of sound could only be recorded on flat surfaces, such as glass plates that were smoke blackened. Harking back to Young’s observations, Scott de Martinville developed a new medium circa 1857, where he wrapped paper sheets around a cylinder. Thus he could now record a continuous spiral of sound waves for a longer time.

　

An impossible vision

When we think of sound recording today, it goes hand in hand with sound reproduction. Thus it comes as a surprise that there was no technological ability to hear any recorded sound until Edison’s phonograph, Scott de Martinville conceived of an idea far ahead of his time so he can be forgiven for failing to see what is obvious today, while Edison should now be considered the inventor of the second part of the process – reproduced sound, rather than recorded sound.

Scott de Martinville thought his invention could be utilised in a variety of ways. He envisaged it being a method for tuning instruments. Alternatively he thought it could be used as a dictation device, whereby secretaries could learn to read the patterns of sound waves! The latter idea was perhaps influenced by his prior career in printing. It sounds like a bizarre idea today that could only be pulled off by some sort of artificial intelligence, even if it was at all possible to deduce words from sound waves! However, science in the 19th Century was a rich environment of possibility and imagination, as evidenced by Tesla’s dramatic failure to build a much promoted wireless energy transmission system that could provide free power internationally!

Édouard-Léon Scott de Martinville

After a report that questioned the effectiveness of the phonautograph by a French semi-governmental organisation in 1858, Scott de Martinville did little work on his invention until the following year, when he collaborated with Rudolph Koenig, a builder of acoustic instrumentation. Koenig made significant improvements to the invention, and even manufactured phonautographs for commercial use. By 1860 Scott was working with a physicist called Henri Victor Regnault who would further refine the device. This was the period when the phonautograph matured, and a number of successful recordings were made, featuring songs and recitations, which still survive in good condition today.

Unfortunately the promise of commercial success did not materialise to any substantive extent for Scott de Martinville. He seems to have stopped making phonautogram recordings by 1862, and returned to a more conventional career as a Parisian bookseller. Years later, Thomas Edison, the famed American entrepreneur and inventor, would realise the true creative and financial potential of an innovation similar to the phonautograph.

Scott de Martinville’s invention would still be extremely important if it had not directly led to further developments in sound recording. However, it would be an even greater milestone in this respect if Edison had prior knowledge of it before inventing the phonograph. The latter would be a contentious view. Sound historian David Giovannoni claimed that Edison had no prior knowledge of the device.

However, it seems the phonautograph was quite a well known device in the US, and another source asserts that Edison used a phonautograph during 1878, for the measurement of sound, due to a court case. Thus it can be suggested that the phonautograph may have represented some level of inspiration for the phonograph. This is a claim that has some resonance because Edison’s invention of the phonograph is noted for being extremely sudden, where a working model was built based only on a single drawing! Ultimately however, well sourced research is required to clarify if Edison had any substantive contact with the phonautograph at a prior time.

　

Musique antique

As an invention, the phonautograph was seen as a footnote in history but in 2008 several American sound historians discovered some recordings in Paris, and attempted to convert them into sound. Sound archivist, David Giovannoni, led the research effort, and a method that could read safely read these precious recordings was devised.

The paper sheets were scanned at very high resolutions and then the sound waves were read by computer to reproduce the sound. Most were in poor condition but they found one pristine phonoautogram recorded on the 9th April 1860 – a ten second segment [download] of a French folk song, Au Clair de la Lune, which translates as By the light of the moon.

 

It is quite an experience to hear the sound of the recording. As would be expected, it is extremely muffled and fluctuates a great deal but it can still be identified as a human voice. Many comments on the Internet featured words like “creepy” and “spooky”, and indeed the recording does have a ghostly disturbing quality. It does seem that many ancient recordings possess an eerie voice-from-the-grave quality but on Au Clair de la Lune it is especially apparent, a lightly pitched voice suggesting that of a girl, a fragile quivering voice, malformed by overt distortion.

It is important to note that the recording would have been effectively inaudible if it wasn’t cleaned up to a huge extent. It was split into 16 parallel sound streams, adjusted, and edited back together to account for speed variation.

A half speed version [download] was released the following year, indicating the source of the voice was a man, rather than that of a girl as heard on the initial version. It is thought the voice could be Scott de Martinville himself. Although there is no absolute reference to compare the recording to, it appears that the half speed version is correct because his other recordings also appear to have been recorded at this speed. Scott de Martinville also recorded tuning fork tones as a reference point for these recordings. Based on his writings, the pitch was originally thought to be 500 Hertz but it is now believed his understanding of the pitch was actually 250 Hertz, as he had his own peculiar method for measuring frequencies.

　

 

When exactly was sound recording invented?

As already stated, Scott de Martinville made numerous phonautograms before 1860 that still exist so it becomes difficult to assert when the first sound recording was actually made, a point where we can say sound recording was truly invented. It is an issue relating to a rather momentous historic event. One could argue that the invention itself dates to 1852 because Scott de Martinville apparently has his “eureka” moment of inspiration that year. Here however, ideas are clearly not sufficient. It could also be argued that the earliest recordings, dating back to 1853/4, represent the moment of innovation, especially since they represent the start of the practical developmental path toward Au Clair de la Lune.

However, the original content of those recordings cannot be reproduced as they produce no identifiable sounds. Similarly the audibility of the later cylinder recordings [download] from 1857 is not apparent, even after audio restoration. They lack any meaningful hint of coherent sound, and they are unlikely to ever be reproduced adequately, judging by the accounts of the archivists. At an essential level, a recording cannot truly be said to be a recording if the event it supposedly stored cannot be reproduced in any form whatsoever.

Since science relies on what is evident and provable, the question in this case should perhaps be revised as “what constitutes the earliest recorded sound that is recognisably from a distinct sound-source”. The aforementioned recording of a tuning fork qualifies as the first recognisably recorded sound [download – shown through the process of sound restoration] because it dates to 1859.

However, the recording of Au Clair de la Lune, and subsequent successful recordings from 1860, are of far more complex acoustical waveforms that would be harder for the phonautograph to record adequately. Therefore, it seems reasonable to conclude that sound recording, as we understand it today, became a reality at a point during the years of 1859 and 1860, where Scott de Martinville’s work, aided by Rudolph Koenig, bridged a yawning chasm by making the crucial leap from inaudibility to audibility.