Final Project: Greensleeves with 8-part harmony, epic organ, and bonafide church bells??

“Christmas Eve” painting by J. Hoover & Son, 1878

This project began, like all my projects, with a random voice memo. In winter 2019, I was playing the tune of “Greensleeves” on the piano with some minor 13th chords. It’s hard for me to articulate, but the recording (and the “Greensleeves” folk song in general, which was written anonymously around 1580) inhabits an enchanted liminal space between wintery, nighttime, enchanted vibes and a classic, elegant, timeless English feel. The sense of timelessness and spaciousness really gets me.

Voice memo:

I thought of doing “Greensleeves” because my bigger goal with this project was to create an arrangement of a *classic song* (classic in this case just meaning old and well-known) in Waveform. I gave myself the following guidelines:

1. includes me singing complex stacks of harmony like Jacob Collier, but obviously in my own style
2. pushes me to craft my own synths in Subtractive and 4OSC
3. pushes me to avoid existing samples and record found objects, manipulating unconventional sounds to create beats, synths, etc

Once I decided on “Greensleeves” tune, I set my priorities as (1) first, (2) second, and (3) third most important. This was going to be 1) an a cappella arrangement, 2) undergirded by churchy instrumental sounds, and 3) supported by samples, with vocal recording obviously at the center.

Then I had to decide which lyrics to use. I wasn’t going for medieval love song vibes, and the original words of “Greensleeves” are kinda weird. There was a surprising amount of alternative lyrics. I really wanted to go for something universal/secular-ish, but “The bells, the bells, from the steeple above, / Telling us it’s the season of love” just wasn’t cutting it (sorry, Frank Sinatra). I settled on the original Christian adaptation “What Child Is This?” by William Chatterton Dix (1865). (I opted to change “Good Christians, fear, for sinners here / The silent Word is pleading” to the alternate “The end of fear for all who hear / The silent Word is speaking.”)

For me the Christian Christmas story definitely inhabits that mysterious, shadowy, timeless feeling that I was talking about earlier (the wise men following the star, the shepherds out at night). I liked imagining reverbed, dark organ and choir sounds fitting into that space.

Above all, I love using harmony to color things. I sang the vocally-harmonized equivalent of my voice memo above. The Waveform part of this was easy: stack a bunch of audio tracks.

Unmixed voice sketch of piano improv:

But the musical part was hard. I was so focused on intonation, rhythm, and line that first of all I sacrificed the synchronization of consonants (even in the final recording they’re often out of sync) and second, I couldn’t improvise. I don’t have enough vocal skill yet for that kind of fluidity or flexibility. Improvisation and imprecision did become an important generative tool, however, in the routine that I fell into for each verse and chorus:

1. Brainstorm a list of adjectives to describe the feel of this verse and its dramatic role in the song.
2. Improvise a harmonization on the piano over and over until I like it and put the midi in Waveform
3. Quickly sing lines over the harmony that feel natural (e.g., change the chord inversions to create a smoother bassline, since the bottom notes on the piano version probably jumped all over the place)
4. Change my piano midi accordingly, and take a screenshot of the piano roll as a memory aid for the contour of each vocal part.
5. Sing vocal parts reasonably well. This was actually the easiest part. Confidence is key. I’m not an amazing singer by any means, but everything sounds at least decent with some brave intentionality. Also, I found that doubling or tripling the bass and melody compensated for minor issues with tuning, breath control, or volume.

Vox:

Piano roll (hard to read, but a good memory aid. Notice that the lines in this one are NOT singable yet lol.. like look at the top line):

Meanwhile, I explored samples and synth sounds that I could use. I tried to record my keychain to get a sprinkley/shimmery sound, but after twiddling for an hour with delay effects and feedback I just couldn’t get the sound I wanted, and used a preexisting sample. Other attempts were more successful: I tapped my stainless steel water bottle with my finger, producing a warm tenor-range tone reminiscent of bells.

I added light reverb, high-shelf-EQ’ed out the high frequencies to reduce the potential for dissonance, and automated pitch-shifting to tune this with the wordless “doodah” thing you heard me sing above (fun fact, my water bottle is exactly a quarter-tone sharp from 440Hz tuning). This effect was very spooky and breathy and therefore I made this section my refrain.

Those bell harmonics were SO COOL, dude (listen close: it’s a major ninth chord in root position, I kid you not).

Original sample:

Pitched, reverbed:

Then there was the organ. Ahhhhh, the organ. I tried to make it myself, I researched crazy additive synthesis stuff with sawtooth waves, and then I discovered that 4OSC had an organ patch that blew my attempts out of the water. Its Achilles’ heel, however, was so darn annoying. The organ patch emanated a high-frequency (2-10kHz) crackle that MOVED AROUND depending on what notes you were playing and got really bad when I played huge organ polychords (which, if you haven’t noticed, is kinda my musical M.O.). My best solution was to automate notch filters that attacked the crackle optimally for each chunk of music. I spent a lot of time deep inside that organ dragging automation lines controlling the cutoff to the elusive median frequency that best subdued the dEmoNic crAcKLe (band name list, Mark?). It also helped in certain sections to automate the Q and gain of various notches and curves, not just for the crackle but also for the overall brightness/darkness that I wanted.

The original organ sound on the final chorus, without EQ (or light reverb/compression):

EQ automation:

Two more very interesting sounds I found. Firstly, in Subtractive, there’s a patch called “Heaven Pipe Organ” that sounds very little like a pipe organ but clearly had the potential essence of a spooky, ghostly, high-frequency vibe to add to my vocals. There was even this weird artifact created by a LFO attached to pitch modulation that caused random flurries of lower-frequency notes to beep and boop. I mostly got rid of it but not entirely, because I was almost going for wind-like chaos. The main things I did were:

1. Gentle compression
2. Reverb with bigger size (to stay back in the mix) but quicker decay (I didn’t want the harmonic equivalent of mud 🤷🏼‍♂️)
3. Quicker attack on the envelopes (I wanted it to whoosh in, but not like 20 seconds behind the vocal because that would be dumb)
4. EQ. I wanted those high freqs to really shine: 

Honestly, that wasn’t a whole lot of change to produce a dramatically altered sound.

“Heaven Pipe Organ” before:

After:

Finally, bells. Christmas = bells. You might recall my post called “Boys ‘n’ Bells” (or something, idr) about Jonathan Harvey’s Mortuos Plango, Vivos Voco. I was seriously inspired by the way he has a million dope bell sounds surround you octophonically. I couldn’t quite do that, but I found some wonderful church bell samples from Konstanz, Germany, (credit: ChurchBellKonstanz.wav by edsward) that I hard-panned left and right during the epic final chorus and mainly tuned to the bassline:

I also found a nice mini bells sound (the 4OSC “Warm bells” patch), which had good bell harmonics preloaded and just needed slight EQ to reduce some nasty 10kHz buzz. I had it accompany the melody of the final chorus, but my favorite use was a sort of “musical thread” that bounces back between a countermelody and a repeated note (I swear to gosh that there’s a specific term for this technique that I learned about in Counterpoint class… oh well):

As I started recording the verses and choruses and filled in the synths/samples, I started focusing more on the overall flow of the piece. Most of the work was cut out for me. There’s a musical antecedent/consequent (call/response) structure within each verse and chorus. This is clearly reflected in the melody and harmony. They start out feeling like a question and ends on a solid cadence. As the arranger, I played around a lot with this structure. Often I flipped the harmonic structure of verses so that they started out stable and then went on extended forays into instability/modulation in order to create a build into the next section. Here’s an example where the second chorus actually builds into the third verse, which surprises you because, well, you’ll see:

What you just heard is probably my favorite part. The second chorus is embarking on an elaborate quest to escape the key of f#minor. A dramatic descending circle of fifths modulation ends on a classic II-V that we have been *culturally indoctrinated* to expect to go to the I (in this case E major). Even going to the vi would be equally unsurprising (vi, C#min, is closely related to E major… this is a standard deceptive cadence). BUT NO. In a moment of serendipitous experimentation, I accidentally dragged an organ midi recording of Verse 3 right after the Chorus 2 modulation I described. The Verse 3 organ midi happened to be in B minor, and the transition sounded BEAUTIFUL. It has those haunting English vibes, and it works harmonically because the voice leading from B major (V in E) to B minor works even though functionally this is nonstandard.

Here’s what the transition from Chorus 2 to Verse 3 would have sounded like if it was predictable (Bmaj to C#min):

Here, again, is the colorful shift I chose from Bmaj to Bmin:

I’m pretty convinced about the second one., but please let me know which one you think is more dramatic in the comments!

The last element of this project I wanna talk a bit about is mixing. Mixing was hard. Because this is centrally an a cappella project, I put the voice in the center (spatially, spectrally, volume-wise). My first main focus was balance of the vocal stacks good. I spent hours arranging the tracks by color (red=melody, blue=bass, in-between = related/supporting parts), putting each separately-recorded chunk into folders within my vox submix, and getting the levels/panning right in Mixer:

For a while, I wanted the emphasize reverb in the vox, but inevitably that causes the vox to sit back in the mix. It needed to be front & center. So instead, I generally tried to compress the vocals for a close/direct sound and made a valiant effort to keep all the other sounds out of the way spectrally. But sometimes I liked the effect of blending a high-freq pad with the vox (see the Chorus 2 demo above) or giving a kick in the bass (see the use of bells). Speaking of bass: I showed my mix to my friend Liam and he suggested boosting the bass to make the vox richer. I did and it helped. Shoutout to Liam.

So yeah. I think that covers a lot of my decision making! Here is the final product. Enjoy, and HAPPY HOLIDAYS!!