|Picture by Brenderous, used under Creative Commons license|
In the course of the research, I suggested that Melodyne, an audio tool that is capable of transcribing audio files to musical notation, could be used to consider the "musicality" of a poet's performance of a work. In particular, I suggested that it offers a unique visualization of pitch dynamics and tempo accelerations and decelerations. As an example, I exported William Carlos Williams reading his famous poem "To Elsie" to a musical score and compared the musical visualization of a segment with the poem's form and content. As I was learning to do this, I also experimented with exporting Williams' voice to MIDI files, which I then read back into MIDI digital instruments and rendered the literal musicality (a term so often used to speak metaphorically of the mellifluence of a poem) of the performance.
I originally did this with a lesser-known Williams poem called "The Wind Increases," read in 1942 for the National Council of Teachers of English's Contemporary Poets Series. I chose it because I found it to be sonically beautiful (in addition to being inspiringly Whitmanian). Here is the recording, taken from PennSound and with some hiss removed to aid in the conversion. Listen to the sonic properties--consider how Williams' performance kind of crescendos in tempo, intensity, and pitch. It's a great example of a poem's aural properties being particularly expressive of its content, I think.
My first step was to convert the poem to MIDI, a format that can be read by digital musical instruments. Melodyne offers three ways to export MIDI: 1) percussive, 2) melodic, and 3) polyphonous. A percussive export offers a single note or tone in the cadence of the source. So for example, when used with speech, it should resemble the prosodic properties of the speech. A melodic export offers the pitch and tonality of the source, too. So it can output multiple notes and resemble the pitch AND prosody of the source. A polyphonous export is like a melodic export, except that it's capable of representing chords, multiple notes hit at the same time.
So let's see what happened when I created a melodic export of Williams' voice from his reading "The Wind Increases" and passed it through an electric guitar MIDI instrument:
As you can see, we are presented with a literal interpretation of the musicality of the poem, give or take (I can't confirm that Melodyne's transcription is error free). Pretty cool, I thought!
Or how about the polyphonous export passed through a drum kit MIDI instrument:
(I also created a bass track using a percussive export, but that's not very interesting on its own.)
All of this brings us to the main event of our show this evening, to the headlining act: The William Carlos Williams Band. Here you have WCW on vocals (reading his poem), guitar, bass, and drums, all at the same time. All of the tracks are different renderings of his voice reading the poem, overlaid:
There you have it. Now you know what happens when you turn someone who knows just enough to be dangerous loose in a poetry archive with sophisticated audio software and too much time!
I'm still considering how, if at all, Melodyne can support phonotextual analysis, but I thought you might enjoy these byproducts of my research.