Transcribing Kendrick Lamar

There is a lot going on in “DUCKWORTH”, between the story, the samples, and the production. I’m just focused on Kendrick’s flow for now, but there is a mountain of musicological study to be done with the other aspects of the song, and how the song relates to the rest of the album.

Check out this sample breakdown by Tracklib.

I begin my transcription where the main beat comes in, at 0:38. So far, I have only done the first twenty-four bars, because even that was a lot of work. I may do the rest eventually. As when I transcribed Missy Elliott’s “Get Ur Freak On,” I have access to the acapella version of Kendrick’s song, so I could use Melodyne to assist my ear. The colors indicate rhymes.

(Here is the MIDI-only version, if you want to hear that for some reason.)

To transcribe the verse, I worked one beat at a time. I looped each quarter note and figured out the syllables’ rhythms, and wrote them out on the chart. Once the rhythms were in place, I listened for the pitch of each note, comparing the results from Melodyne with what I was hearing myself. Once I did the whole verse this way, I checked my transcription’s accuracy by exporting the MIDI from Noteflight and bringing it back into Ableton. Then I listened through it a phrase at a time, comparing it to the acapella, listening for places where my melodicization didn’t align, and adjusting the score accordingly.

Kofi Agawu warns that Western notation is not always adequate to the task of representing African or Afrodiasporic musics. However, he also recognizes that sometimes it’s the best available tool, so we may need to go ahead and use it, being mindful of its limitations. I stand by my transcription’s accuracy, but you should read it as an abstraction of Kendrick’s performance, not as precise documentation. Consider it an assistive tool for listening to the recording, not an authoritative statement on what Kendrick is “really” doing.

My chart is as objective as I can make it, but there is still plenty of interpretation involved. Even though I can see the rhythms from the waveforms in Ableton and the pitches from the graph in Melodyne, I still have to exercise judgment in deciding how finely or coarsely I should be quantizing everything. I assume that Kendrick defaults to a sixteenth-note grid with occasional thirty-second notes and triplets. But there are several places where he uses unambiguous quintuplets, too; I talk about those below.

I exercised even more interpretive judgment with the pitches. On this verse, Kendrick’s pitches lie mostly within the fourth between A3 and D4. I could have transcribed his flow using general pitch zones as per Robert Komaniecki and this 12Tone video: “high” for the syllables on D and C-sharp/D-flat, “low” for the ones on A and A-sharp/B-flat, and “middle” for B and C. But that would not have been specific enough. In measure five, B functions as a “middle” pitch, but in the next measure, it functions as a “low” one. In a way, I did use the pitch zone model, but I treated every chromatic pitch as its own narrow “zone.” To get a sense of how much editorial judgment was required in mapping this verse to the piano keys, take a look at the first two bars in Melodyne, which I’ve annotated with the lyrics.

The orange lines show Kendrick’s actual pitch. Not all of these tiny squiggles are as salient as others. Melodyne has made its own algorithmic judgment as to which are the “real” notes, which it shows as orange blobs. You can see how much abstraction it applies to the first word, “Oh.” In this single syllable, Kendrick starts on B, leaps up to D, wavers around, dips to D-flat, and then ends on E-flat. Melodyne guesses that the main pitch here is D, and my ear agrees, so that’s what I wrote. In general, I tried to place each syllable on a single pitch. The word “hard” is clearly split across C and B, and I notated it as such, but I didn’t do many of these splits in the interest of readability.

I love the clarity and objectivity of the Melodyne visualization, and I could diagram out the whole verse this way, but there are two problems: it’s very labor-intensive to annotate it by hand, and the result isn’t interactive and auditory the way the Noteflight chart is. So I’ll continue using Noteflight until I find something better.

Let’s look at the musical aspect of Kendrick’s flow. Mitchell Ohriner distinguishes between “music-rhythmic” flow and “speech-rhythmic” flow. The idea is that music-rhythmic flow aligns closely with the metrical grid, while speech-rhythmic flow uses freer time and is independent of the metric grid. Kendrick uses music-rhythmic flow on “King Kunta” and speech-rhythmic flow on the later verses in “Momma.”

Ohriner thinks you should notate music-rhythmic flow on the sixteenth-note grid as much as possible, rather than trying to literally represent every microrhythmic discrepancy. For example, he doesn’t think you should write triplets unless it’s in the context of a long stretch of Atlanta-style triplet flow. Ohriner reasonably asks, once you start doing a few syllables in triplets, why stop there? Why not quantize everything arbitrarily precisely using tiny note values and nested tuplets? That’s what Martin Connor does in his transcriptions: he loads them up with tuplets and microrhythms. Ohriner thinks this removes as much clarity as it adds.

For me, the question is not “What is the most precise quantization possible?”; by increasing the cardinality of the metric space, arbitrary precision is possible. The question is “What is the metric structure the emcee presents in performance?” While we might not have a definitive answer to this question in every case, the discourse of emcees like Rakim makes cardinality-16 structures especially likely. Furthermore, with quantizations of arbitrary precision, we can no longer speak of emcee’s delivery as “ahead of” or “behind” the beat (para [2.4]).

I decided to split the difference between Ohriner’s and Connor’s approaches. I default to the sixteenth note grid, but I use more precision when it seems like Kendrick is subdividing more finely on purpose.

Ohriner raises the fascinating question of whether there is such a thing as “consonant” and “dissonant” rap. In “Momma,” he sees the music-rhythmic (on-beat) parts as functioning like consonance, and the speech-rhythmic (off-beat) parts as functioning like dissonance. I would believe this interpretation for “DUCKWORTH” too. There is an uneasiness to the track, reinforced by the woozy tuning of the bass and the jumpy edits between sections, so a “dissonant” flow suits the mood. Kendrick never uses fully speech-like flow in this verse, but he does push and pull hard against the grid, making it feel like he’s speeding up and slowing down. For example, on the line, “You got to love him, you got to trust him, I might be buggin’,” the three phrases have the same rhythmic and pitch contour, but “I might be buggin'” is slowed down.

Several of Kendrick’s rushed phrases create quintuplets. He does so much of this rhythm that I have to believe he’s using it with deliberate structural intent, and that the quintuplets aren’t just the result of expressive deviations from the grid. Take a look at measures seven and eight, the lines “infomercials and no sleep, introverted by my thoughts, children listen it gets deep.” Each of these three phrases is a quintuplet followed by a pair of sixteenth notes. That has got to be deliberate.

Kendrick isn’t the only emcee who uses quintuplet flow. Noname uses it on her (very speech-like) track “Don’t Forget About Me.” Talib Kweli uses it on “Redefinition” by Black Star; Busta Rhymes uses it on “Holla“; and Madvillain uses it on “All Caps,” as analyzed by Kyle Adams. How do you learn to internalize quintuplets deeply enough to rap with them? I haven’t seen any practice techniques for emcees, but there are plenty of tutorials for drummers. Adam Neely explains how to get a “drunken drummer” feel using quintuplets (and other tuplet groupings.)

Slynk shows how you can create quintuplets in Ableton Live. I learned many challenging rhythms by programming them first; maybe that approach will work with quintuplets too.

It occurred to me that I don’t even know how to count quintuplets. Someone on Reddit says you should count them “One E and A ka Two E and A ka…” (by analogy to counting sixteenth notes as “One E and A Two E and A…”) Another Redditor suggests mnemonics: thinking of “geophysical,” “hippopotamus,” “inaccessible” or “university” for 2+3 groupings, and “Constantinople,” “communication,” or “mummification” for 3+2 groupings.

My original goal in creating my rap transcriptions was to give my music education students a foothold into hip-hop creativity. The thing is, I don’t want them to learn and perform these songs. Instead, I want them to follow a songwriting approach I learned from Toni Blackman: to write their own lyrics for existing rap flows. I don’t expect any novice student of rap to be able to deliver a flow like Kendrick’s; the point of close listening to him is to show the students how complex and technically demanding rap can be. More broadly, I’m hoping to direct more musicological interest toward artists like Kendrick, not because he needs academic validation, but because so many music academics would benefit from close engagement with hip-hop.

Update: this post is the subject of some intense and valuable debate on the Hip-Hop Music Ed group on Facebook.