Here’s a new Claude McKay poem song setting, “To a Poet,” completed as part of my concentration on his poetry this February. Somewhat of a “deep cut” in McKay’s poetry, but as sometimes happens when reading a bunch of poems, there was one set of lines that stood out as I read this pioneering Jamaican-American poet’s work. I’ll get to those lines, but first a detour about the music.
Which is something I don’t write much here. The Parlando Project started as a musical idea, though I always thought I’d want to say something about the experience of encountering the poems on the way to making the music, and so this blog. While true engagement with the blog posts here is hard to judge – something made harder by occasional bouts of what appear to be bots skimming (and reskimming) posts here since last autumn – there’s been a satisfying and unrelenting increase in visits to this blog over 10 years. Blogs may no longer be the new-hot, but the visits keep coming. Thank you, readers, and I wish that my personality and situation would allow me to be more attentive to your comments.* I treasure anyone that spends a little time here.
But the musical pieces, the cause of this all? They get no more listens than they did only a few years into this Project when monthly readership and listening numbers were roughly equal – while presently listeners are maybe 10% of the numbers of readers. I’m realistic about the limitations of my musical expression. I’m about as far from a poptomist composer as could be imagined, and I’ve long feared that not sticking to one style of music creates what used to be called by radio programmers “button pushers” – those who hear one or two songs they strongly dislike and decide to go elsewhere. I understand, people react to music sensually, emotionally, and so a bad experience with music creates a stronger distaste than a duff blog post or the choice of a poem some don’t care for – but my music making reflects my listening, it’s eclectic. I’m committed to musical adventurousness and variety, and so by intent or missteps the music may not always be something you’d choose. I don’t believe I’d enjoy this Project if it was anything else.
But one musical constraint has always been with me: my singing voice and its take it or leave it limitations. Spoken word poetry, even with integrated music, is one thing, and I’ll do that, but some words tell me they expect to be sung, and almost always, that means the singer is me. I’m grateful for the times my long-time LYL bandmate Dave Moore has given me and my listeners a break, and there have been scattered other “guest vocalists” over the decade. Today’s song is different. I wanted to do something (however simplified, as I think simplified music retains powers) more like Art Song, the composed music that features trained singers who express strong melodic contours through skilled techniques, rather than off-the-cuff, I’ll-give-it-a-go, folk-singing.
Go ahead, scroll down and listen to today’s piece setting to music the words of Claude McKay. I’ll wait. And here’s the text of McKay’s poem that was performed with my music.
.
OK, you’re back. Who’s the guest vocalist? The stage name is Felicia. Felicia is a virtual instrument. Readers here may recall what that is, but I think it’s important to reiterate: a VI is often a set of atomized recorded samples of a real “acoustic” instrument, a whole range of notes with articulations expressing a range of colors. Let’s not confuse two-letter-acronyms, VI is not AI as expressed by services like Suno which take an overall text description of the nature of a song and create in one fell swoop the finished melodic lines of the singer (and all their arrangement and accompaniment) by apparently recombining conventional musical materials. If I take a VI of, for example an organ, and invoke it on my computer it won’t play a toccata, early Phillip Glass, or “Rock Lobster,” or anything like those pieces, no matter how I write a run-on requesting sentence about it. It instead needs me to play my MIDI guitar or little plastic keyboard, or inscribe notes on a MIDI piano roll notation. I could (as a very limited keyboardist) invoke arpeggiators or chord and pattern generators to extend what I play or write – but it still feels like composition to me. In contrast, AI like Suno feels like I’m a royal patron asking my musicians for some conventional musical noise to underscore my cultural pretenses, caring only that it be inoffensive.
Another difference: even a despot likely paid their court musicians something. AI? Not so much. The company that sells the Felicia voice VI (Dreamtronics) claims it pays the human singers that it samples to make its products.
Still, I expect this revelation, or its implementation in this piece, will repel some listeners. The spirals inside the ear have their own Uncanny Valley, and I too feel that fearful symmetry. It sounds like a human putting artifice onto itself or an artifice taunting the qualities of human. I put up with it because I had a goal, and my voice could never sing with the VI’s technique, particularly over a longer piece. I played the vocal line on my little plastic keyboard, and typed each syllable as text that the notes would sing. Early vocal VIs were very picky about needing explicit phonetic text, but this one knows much more about the baroque tangles of English pronunciation, yet I still had to tweak some syllables. The program has a range of controls for expression variations – learning how to use them will improve results – but it presents default expressive choices that keep the monotonous spiel of old-school robot speech away.**
Felicia made it much easier to get something that wasn’t fakey bad or unintelligible as older vocal VIs I’ve tried, even as I still felt the need to do work to improve problematic passages. It took me several hours to create the vocal line realization you hear in today’s piece. A trained human singer at the mic could have done it in an hour, even including leeway for retakes and “try it this ways.”
One thing I noticed: even when I had polished up the intelligibility of the VI sung text, the meaning of the words seemed abstracted to me as a listener. Oddly, this is the same thing I sometimes hear as a listener with Art Song, where the composer’s elaborate melodies or the singer’s concentration on demonstrating virtuoso technique make the words vehicles for expressing music more than shared experience. Human vocalists singing Art Song, in their own way, produce their own unsettling Uncanny Valley.
Which may bring up the question, why not just use a real singer? Yes, that would be better. I, who am socially awkward, not able to schedule a time to do my creative work, and heading-up a non-revenue Project can say that would be an ideal, but unlikely, option. Furthermore, in the process of composition I wouldn’t be able to test my musical choices as a limited singer, so there’d likely never be a score for such a singer to follow.***
Will I use a vocal VI again here? Likely, though I don’t think I’ll use it most of the time, or even often. My voice is my voice, and I feel I should use it. Still, I was very happy that I could realize this musical piece. Through the technology of VIs I was also able to play the atomic recombinations of an oboe, a viola da gamba, and a hurdy-gurdy as part of the accompaniment.
OK, so back to Claude McKay’s poem. Like his “When I Have Passed Away” from earlier this month, McKay’s poem here speaks of posthumous poetic legacy. In the context of this month while using a computer VI to sing his words, I was much taken with the pair of lines in his poem: “The modern kings will throttle you to greet/The piping voice of artificial birds.” Maybe 100-years-ago, a prophetic McKay knew how I’d come to try to make a song of his poem using a voice of artifice under the rule of a disordered king and assorted technological barons.
You’ve already heard that song haven’t you? What, you didn’t obey the words and stop partway for the music? You’re a rebel! An outlaw! A traitor! You have one – no, two – more chances to hear my musical setting of “To a Poet:” with the audio player you should see below, or with this highlighted link that will open a new tab with its own audio player.
*Besides mundane life-chores, these increasing problems are partly from aging, and partly secondary to the folks I share my home and life with. It’s hard for me to devote regular and predictable time to this Project, and when opportunity time comes, my nature is to work on finding new words and creating the musical pieces. For some reason (aging? self-doubt?) it’s increasingly hard for me to make the social small-talk that should be trivial and expected.
**I doubt this variation of expression is heavy duty AI either. For a long time VIs have used a pseudo-random cycle of expression variations as an option in their programs. I didn’t get a sense the VI generally knew from the denotative sense of a word or placement in a sentence or musical phrase to give it a particular kind of invocation.
***A product like this, it seems to me, would be ideal for a composer who doesn’t sing well and who would like to rough out scores before an actual performance, as it might give a better quick approximation than just playing the vocal line’s notes on a piano for instance, just as orchestral composers are increasingly roughing out arrangements on other virtual instruments.