Attempt at a Definition of “Sound Recording”

In my article about Hercule Florence, Florence “fixes” animal voices without a microphone, without tape, without a disc. He listens, then he writes. And a question slipped in between two silences: does zoophony belong in my dossier on the history of recorded sound, or is it already something else?

To move forward, I tried to build myself a simple definition, almost a field definition. Florence, in zoophony, does not capture a sound: he translates it. He turns a sonic experience into signs, pitches, durations, rhythms. In other words, he creates a score. In my own article, the idea is very clear: it is a “fixation of sound through writing,” an archive, but an archive made of musical language, not of sonic material.

And that’s where the word recording starts to buck. A recording, originally, is a trace caused directly by sound, a physical link between the event and its imprint. Édouard-Léon Scott de Martinville’s phonautograph, for example, “writes sound” as a line. At the time, you can’t listen to it, but the sound did guide the stylus, with no interpreter in between. It is completely graphic, but it is already a capture.

Zoophony, on the other hand, adds a human link, and therefore a margin: the ear chooses, sorts, simplifies, cleans up. Like when Olivier Messiaen notates birdsong: he does not “replay” the bird’s song, he recomposes it. Even when he relies on recordings, which happens in certain pieces, the final result remains musical writing, therefore a reinterpretation. The bird becomes material, not an imprint.

So where do we draw the line? Here is a very practical proposal, meant to help classify without dismissing: a sound recording is the fixation of a sonic event in the form of a trace, analog or digital, enabling later playback (even if imperfect), without going through a symbolic translation. A sound transcription, by contrast, is a fixation in the form of symbols (notation, text, onomatopoeia, scores), which always requires a performer to become sound again.

You might think a digital recording is also symbolic writing, since everything ends up as 0s and 1s. But the decisive difference is that these symbols are not a human translation of sound, like a score. As I point out in my dossier on digital audio: they are numbers produced by a physical measurement of the signal. When we digitize, we sample the sound’s amplitude at regular intervals and quantize it into values, then encode those values in binary to store and transmit them. The 0s and 1s do not describe the song the way a musician would; they encode a measured trace that can be converted back into a waveform by a digital-to-analog converter, without artistic interpretation. In other words, yes, it is a form of symbols, but machine symbols and reversible ones, designed to reproduce sound almost perfectly, whereas zoophony (or Messiaen’s notation) always imposes a reading, choices, and therefore a reinterpretation.

Fine, “artistic interpretation” may separate a transcription from a recording, but what about technical interpretation: the one imposed by all the gear between the source and playback? And what about the listener’s human interpretation, shaped by education, culture, and listening habits? What do you think?