Back in 1857, a man named Édouard-Léon Scott de Martinville developed and patented a device that he named the phonoautograph.  The basic structure was as follows.  A strip of paper was blackened with smoke, and then rolled slowly as a person spoke into a cylinder like vessel in which there was connected a sound sensitive membrane attached to a needle.  As the person spoke, the needle would vibrate and trace patterns into the paper that matched the wavelengths and amplitude of the recorded sound.

Here is an early version of the machine:

Here is a detailed scan of what one such phono autograph strip would look like.

At the time, this invention was somewhat of a miracle, as it essentially worked to empirically reveal the way in which sound manifests itself in a visual format.  It took one sensory phenomenon and wrenched it out of its standard milieu, forcing it into a transcribed form that then, in an almost alchemical fashion, presented sound forever captured into the medium of writing.  But given the technology of the time period, this is effectively a memento mori.  In order for orality/aurality to be rendered visible, it must be killed, insofar that the only remaining evidence is the trace of where it once resided.  What defines sound in certain theoretical registers is its inherent ephemerality, wherein the principle experience of encountering the sound is in and of itself an instantaneous transmission and one that is destined to evaporate.  Granted, in the contemporary moment wherein recording media come attached almost as an afterthought smart phones, this type of logic may lose a bit of its weight, however the spirit is still there (or not, all things considered).  The primary transmission of sound, in this case the voice, retains its instantaneous emergence and disappearance, its simultaneous presence and lack thereof, and as such even a FLAC file of a voice remains a pseudo-shell of the origin.  A hi-fi crypt of the impulse towards communication, in a sense.

And there is even some interest in the skeuomorphic tendency of some of these programs.  The voice memo function of the modern iPhone traces out wavelength and amplitude as it records, calling into mind the spectral function of the phonoautograph.  As it records even the most quotidian of files, the trace develops before the eye, preserving and etching the epitaph of that moment in simultaneity.  Lots of talk about death here, but it is apropos considering the direction in which this will lead.  At the time of its invention, the phonoautograph was concerned only with visually etching the concept of sound; pinning the butterfly, if you will, and silencing the beat of its wings.  However, technology never ceases to break open crypts and reanimate corpses long since dead, and in 2008, spectrographic sound programs allowed just that.

Reverse engineering the process of the phonoautograph, sound technicians were able to use some of the strips which were most intact and use the data within to reconstruct them into sound files.  Here is the earliest recording of the human voice, from 1860:

http://www.youtube.com/watch?v=mg2s_A3stGM

There is something inherently disturbing to me about this sound file.  In fact, there are several things.  First and foremost, the quality of the sounds itself, clipped and muffled, does no favors to reduce the uncanny nature of the recording.  As Derrida frequently insists, there is an impulse on the nature of humans to attempt to achieve purity in essentially all aspects of culture and society, particularly in the realm of experience.  Even the most lo-fi of recordings (hello, Ariel Pink), can be accepted if an intention towards the art form can be deduced, however broken sound files or incomplete ones maintain an instant aura of mystery if not outright uncanniness for many.  It is as if the sound file violates the ideal experience of human expression, despite the fact that if one follows the line of Derrida, such an idea of pure presence is broken from the start.  There is a clear melodic structure to the sound which can be easily deduced as the song "Au clair de la lune", and as such, a basic "humanness" can be sensed.  However it is one that is shattered, distorted, and muffled almost entirely beyond comprehensibility.  It is for me, for all intents and purposes, how I would imagine the language of the dead.

And is this not precisely what is happening here?  Literally speaking, of course.  However, this idea extends beyond the confines of mere temporality.  The discovery of this recording shattered previous conceptions of the first known recording of the human voice, that being Thomas Edison's in 1878.  Granted, this isn't at all that much later than the recording of Martinville's voice, however the differing intention between the two causes a bizarre warping of temporality to occur.  Edison's recoding, intended to be heard, set off the next great leap in media technology towards the preservation and replicability of the human voice via sound media.  Dead voices, to be sure, however ones which find themselves enclosed within their intended burial site.  The phonoautograph recordings of Martinville, however, feel like the specter of a forgotten group released from their grave centuries later.  It took one hundred and thirty years for these voices to finally be heard, so in this sense, the mystique and uncanny power has gestated for this long as well.  This isn't the vanguard voice of Edison, opening the channels for new types of history with one loud utterance.  The voice of Martinville is a specter doubled over, the specter of a spectral inscription that, in its original intention, was never meant to exist.  This is of a person sitting in a room wanting to look at what his voice looked like; a sonic alchemist indeed.  And now, we can finally hear it, in all of its phantasmic and garbled beauty.  

That both of these firsts, Edison and Martinville's, happen to both be songs is worth consideration on its own.