NOTE: This initial account of methods for transducing images into sound—first posted in 2009—is maintained here for its historical interest. You can also find it archived in Archive.org's Wayback Machine. For current information, please visit griffonage.com.
2.1—We can use the same technique to play earlier inscriptions in the same format, even if they were not produced automatically using a sound spectrograph.
Take, for example, the manually produced graph of "Joe took father's shoe bench out" published in John C. Steinberg, “Application of Sound Measuring Instruments to the Study of Phonetic Problems,” Journal of the Acoustical Society of America 6:1 (July 1934), 16-24, at page 23:
This image requires some graphic processing before we can play it, apart from the matter of inverting it from black-on-white to white-on-black. Specifically, it has a grid overlaid onto the marks that correspond to sound, and we need to remove the grid; otherwise we'll hear it too. I will refer to this process as degridding. Here's what the inscription looks like afterwards:
The frequency scale is logarithmic and runs, as nearly as I'm able to estimate, from 62.5 Hz at the bottom to 5000 Hz at the top. The duration is 1.5 seconds.
So what does it sound like when we run it through reverse spectrographic software? Here's the answer:
2.2—Here's a second example of the same sentence, this one from Harvey Fletcher, "Some Physical Characteristics of Speech and Music," Journal of the Acoustical Society of America 3:2B (Oct. 1931), 1-25, at page 3, played with the frequency range set to 62.5 to 8000 Hz logarithmic and the time to 1.5 seconds.
2.3—The same format was also used to show melodic pitch contours, particularly in the work of Milton Metfessel and his colleagues working under Carl Seashore at the University of Iowa. Metfessel's technique, which he called phonophotography, consisted of recording sound as a waveform on strips of photographic film and then measuring and plotting the pitch contours by hand. One of his most elaborate inscriptions, published in Phonophotography in Folk Music: American Negro Songs in New Notation (Chapel Hill: University of North Carolina Press, 1929), 82-84, shows the four parts of an African American vocal quartet singing "Swing Low, Sweet Chariot." Metfessel recorded each of the four parts separately and later plotted them together on a single graph using different colors. The song was spread out over five separate plates; here's number two:
Listen to Metfessel's phonophotographic inscription of "Swing Low, Sweet Chariot" played back in stereo, with the red lines in the left channel:
The playback is set to approximately one bar per second. Note that I've had to reconstruct the lines where they are obscured by the heads of musical notes. Metfessel inserted the rectangles with regular diagonal lines as needed to extend individual parts so that the whole would line up.
2.4—Together, the graphs of "Joe took father's shoe bench out" (1931, 1934) and Metfessel's "Swing Low, Sweet Chariot" (1929) demonstrate that reverse Fourier analysis software can play inscriptions made before the advent of Bell Laboratories' sound spectrograph, and that it can do so meaningfully, automatically converting the images not just into sound, but into the specific sounds their creators intended them to represent.
