Reconsidering Earliest-Known Recording Audio historians David Giovannoni and Patrick Feaster discuss a revision to their discovery last year of the earliest-known recorded sound from 1860. They have determined it was being played twice as fast as it needed to be.

Reconsidering Earliest-Known Recording

  • Download
  • <iframe src="" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript


This is ALL THINGS CONSIDERED from NPR News. I'm Melissa Block.


They are audio historian David Giovannoni and Patrick Feaster of Indiana University. Welcome to both of you.

DAVID GIOVANNONI: Thank you, Robert.

SIEGEL: Is it fair to say that you've concluded A, last year's recording is as old as you thought, but it's not exactly what you thought and that you've also found a set of recordings that are even older than that one, Patrick Feaster?

PATRICK FEASTER: That's right. We've decided that instead of playing back the voice of a young girl, possibly the daughter of the inventor, that now we're actually hearing the voice of the inventor himself. We played it back at the wrong speed. But in addition to that recording, we've stumbled across a trove of new materials at an archive in France dating from three years earlier, so a batch of recordings from the year 1857.

SIEGEL: Now, let's begin with point one. The recording from last year was made by a man named Edouard-Leon Scott de Martinville?

FEASTER: That's right. Edouard-Leon Scott de Martinville was the inventor of the idea of recording sounds from the air using a membrane. So the basic idea is that covers all microphones, all sound recording devices that we have today.

SIEGEL: Unidentified Man: (Singing in French)


SIEGEL: David Giovannoni, it sounds like a young person or a high-voiced person singing the first line of the French song "Au Clair de la Lune."

GIOVANNONI: That's right. That's what we thought last year. It sounded natural to us. What we've learned since is that we played it back too fast.

SIEGEL: Unidentified Man: (Singing in French)


SIEGEL: No, my question is how do you know that that's the right speed, Patrick?

FEASTER: Well, a few months ago, I went to play back another of the phonautograms in that same group, a recitation in Italian. When I went to play this back at the same speed we'd used for the earlier phonautogram, it came out sounding like The Chipmunks.


SIEGEL: And you concluded there's something unnatural in what we've heard now, the reading in Italian.


FEASTER: Didn't sound quite right. When we played it back at half that speed, it sounded like natural speech.


SIEGEL: Do we know from his notes, by the way, that that is Edouard-Leon Scott de Martinville?

FEASTER: He gave us a clue. There's a footnote that says: I was wrong. The order of the words should be (Speaking in Italian) instead of (Speaking in Italian). By taking...

SIEGEL: That's Italian for in human form or in human shape.

FEASTER: Mm-hmm.

SIEGEL: So we know, then, that he was talking about himself making that recording, and the high-pitched voice would therefore be rather unlikely at the speed from last year. How much does this rearrange the history, so far as we know it, of recorded sound?

GIOVANNONI: It doesn't rearrange one part of it. Right now, this recording of "Au Clair de la Lune" is still from April 9th, 1860. That didn't change. What does change is our interpretation of it. Instead of a window opening up into the past, where this young woman is reaching out to us, singing "Au Clair;" instead now we understand the sounds coming through that window to be that of the inventor himself doing an experiment.

SIEGEL: Now, there are also recordings you've discovered that are older than these 1860 recordings. When do they come from?

GIOVANNONI: This other group of recordings was made by Scott in 1857, the first year he really worked on this idea, and these recordings push back by three years the date at which we have a large batch of his materials.

SIEGEL: Do we get to hear any songs this time or musical scale, something we can - some Italian being read?

GIOVANNONI: I'm afraid not. We can hear hints of speech sounds, but the speed is very irregular.


GIOVANNONI: Because this was being cranked by hand. In the other case, we could correct for that. In this case, we can't.

SIEGEL: And is Mr. Scott de Martinville, is he honored in his native land? Is he a well-remembered and well-thought-of inventor or as obscure there as he is here?

FEASTER: Up until last year, he was...

SIEGEL: Fairly obscure.

FEASTER: Pretty obscure and probably still is, although what we have is truly, as far as we know to date, mankind's earliest recording of its own voice. That's a pretty special thing.

SIEGEL: Well, audio historians David Giovannoni and Patrick Feaster, thanks a lot for talking with us about this.

GIOVANNONI: Thank you, Robert.

FEASTER: Thank you.

Copyright © 2009 NPR. All rights reserved. Visit our website terms of use and permissions pages at for further information.

NPR transcripts are created on a rush deadline by an NPR contractor. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.