Groove melodies

Like harmony, melody works differently in grooves than it does in linear songs or Western classical compositions. In this post, I try to figure out what makes a good groove melody, and how to write one.

Update: Joshua Horowitz made an interactive animation of this image! It’s so cool.

The endless loop

The main requirement for an effective groove melody is that it should sound good if you loop it endlessly. In practice, you might only repeat the melody four or eight times, but it should sound just as good if you repeat it sixteen or thirty-two or sixty-four times. Your groove melody shouldn’t become annoying after all those repetitions; it should be cumulatively more effective, sending you into a hypnotic trance.

The obvious place to look for examples of successful groove melodies in Western music is in Afrodiasporic loop-based musics: blues, jazz, gospel, R&B, country, rock, pop, funk, dance, and hip-hop. However, as I’ll discuss below, digital audio editing also makes it easy to discover good groove melodies in any kind of music at all.

An example: “Manteca”

The head A section from “Manteca” by Dizzy Gillespie, Chano Pozo and Gil Fuller is an all-time great groove melody.

The rhythm of this melody is a mambo bell pattern. If you want something that’s going to work when repeated, rhythmic cells from the African diaspora are always a good starting point. Notice that the riff avoids a sense of harmonic resolution by ending on A-flat, the flat seventh of the key, rather than a stronger note like the tonic B-flat. Ending a phrase on an unexpected note is part of Dizzy Gillespie’s style, but it’s also a common trait in groove melodies, as I discuss below.

Here’s a MIDI version of the A section melody, wrapped in a circle:

The tune is almost a simple repeating cell, but not quite. The third time through the riff is different from the first two times. That gives it a pleasingly recursive call-and-response structure.

• The two notes in the first beat are a call, and the two notes in the second beat are a response.
• The first three notes in the high part are a call, and the fourth and fifth notes are a response. 
• The first half of the first bar is a call, and the second half of the first bar is a response.
• The first bar is a call, and the second bar is a response.
• The first two bars are a call, and the second two bars are a response.
• The first time through the A section melody is a call, and the second time through is a response.

The brackets show all these pairs graphically:

“So What” by Miles Davis is another great groove melody with nested call-and-response structure. I haven’t investigated this idea systematically, but I suspect that any really compelling groove melody will have this fractal-like quality of structure at several different levels of scale simultaneously. If a melody is going to be gratifying after many repetitions rather than annoying, then it has to be deeply interesting, and people do seem to find fractals to be extremely interesting.

Creating groove melodies from samples

You can discover good groove melodies by looping samples of audio in a DAW. Ableton Live is especially well designed for this purpose, but any DAW will do. In his book Making Beats, Joseph Schloss says: “Record collecting is approached [by hip-hop producers] as if potential breaks have been unlooped and hidden randomly throughout the world’s music. It is the producer’s job to find them” (p. 37). The art of sampling is all about identifying short segments of music that are satisfying when you connect their beginning and end together. It’s easiest to find samples in music that was loop-based to begin with. You can throw just about any randomly chosen two-bar segment of any James Brown song into the sampler and get a good result. But it’s more challenging (and often more fun) to find a good loop in a more linear context.

Thelonious Monk is an inexhaustible source of great loops. Try the intro or the opening bars of “Bemsha Swing,” “Thelonious,” “Brilliant Corners” or “Green Chimneys.” Also try literally any two-bar segment of any of his solos. All the jazz greats are rich in potential samples, but Monk’s ideas are particularly well-suited, probably because they’re so fragmented and harmonically idiosyncratic.

Classical music isn’t usually a good source of loops, but the prelude to Bach’s Cello Suite in G Major is a conspicuous exception. Pretty much any phrase from this piece makes a great groove melody when you loop it. You can hear my other classically-based grooves here and here.

Some more standouts from my sample collection that make fantastic loops:

• This Bob Wills trumpet lick
• This McCoy Tyner lick
• This Jerry Garcia lick
• This Vassily Kalinnikov phrase, which contains the saddest chord progression ever
• This equally heartbreaking section of Henry Purcell’s Dido and Aeneas

No harmonic resolution

A good groove melody should avoid a dramatic narrative of harmonic tension and resolution. The melody can have some tension and resolution (it’s hard to avoid it), but you don’t want conclusive-feeling V-I cadences, because they break up the flow. There’s a reason you don’t hear too many leading tones in pop melodies.

Pentatonic scales are a good match for groove-based music because without harmonic dissonance as a structural device, you have to do all your tension and resolution using metrical/rhythmic dissonance, and that fits the groove aesthetic better. But pentatonics aren’t the only way to achieve harmonic ambiguity.

The hook from “Tom’s Diner” by Suzanne Vega is a good example of how a lack of clear resolution keeps a loop feeling open-ended. It’s in F-sharp minor, but it ends on the third, A, which creates an ambiguous sense that it might really be in A major. The last note is on a weak beat, too. It keeps you guessing, and that keeps you listening.

The end groove of “Hey Jude” by the Beatles is another good case study. It’s in F Mixolydian mode–modal harmony and grooves go together hand-in-glove. The phrase ends on C, rather than the F you might naively expect. C is the fifth, still a resolved sound, but not as stable as the root would be. (F Mixolydian has a pleasing ambiguity with C Dorian mode, a fact that rock musicians exploit constantly.) The resolution is further weakened by the fact that the word “Jude” falls on an extremely weak beat, the sixteenth note before the downbeat of the last measure.

Horace Silver’s “Song For My Father” is another iconic groove melody. It’s in F minor. The A section melody ends on G, and the B section melody ends on B-flat, both over the tonic Fm7 chord. These notes would sound very unstable if they weren’t anchored in such a rock-solid groove.

Blues and groove

A lot of Western groove-based music descends from the blues, and blues tonality continues to be an excellent resource for groove melody. Some music theorists explain the seeming disconnect between the melody and harmony in the blues by saying that the relationship doesn’t matter that much, that there’s a “melodic/harmonic divorce.” I don’t love this term, because I do hear a relationship between melody and harmony and the blues, it just operates according to different rules and conventions than Western European tonality. Tension and resolution are still there, but the concepts of consonance and dissonance are very different. Albert Collins’ song “Don’t Lose Your Cool” includes a fascinating example.

The tune is in F blues. Collins’ guitar melody on the first chorus is just the note F. When he gets to the C7 chord at 0:11, he keeps playing F. According to the rules of tonal theory, this is completely unacceptable. The “consonant” notes on C7 are conventionally C, E, G, and B-flat. In a jazzier context, D and A sound nice too. If you were feeling really adventurous, you could even get away with F-sharp. But F? No! It clashes unacceptably against the E.

Or does it? In a blues groove, the blues scale is the universal solvent, and every note works against every chord. If we ever come up with a good theory of the blues, it will probably say that the tension and resolution come from the rhythms rather than the harmonic relationships.

Write from the rhythms up

If you want to write a good groove-based melody, it’s a good idea to start with the rhythms, and work the pitches out later. Or don’t! Keep everything on a single pitch. (Remember that Albert Collins riff you just heard?) Many groove melodies work just fine if you sing/play them on a single pitch, e.g. the tonic. The melodic shape is decoration; the rhythmic placement of the notes is doing the heavy musical lifting. You can hear this in the bassline from “Chameleon” by Herbie Hancock, which sounds great if you play it entirely on the tonic, B-flat:

“It’s About That Time” by Miles Davis is another groove melody that works just fine if you play it on a single pitch.

Remember how “Manteca” is a melodicized version of a mambo bell pattern? Any good groove melody should work well as a percussion part.

Creative approaches

Aside from pulling samples from your DAW, how do you write a good groove melody? My preferred method is to record improvisation to a beat. Then I listen through and hunt for samples, same as if I were digging the crates generally. You can do this by recording your voice or an instrument, but it’s easier to record MIDI via a controller. If you come up with something good, it’s much easier to polish it up, fix unwanted notes, tighten up the rhythm and otherwise mold and adapt your ideas with MIDI than with audio. You can always work out ideas with MIDI and then learn and perform them with your voice or instrument afterwards.

I’m not much of a singer, but I still find my voice to be more intuitive than any instrument. I’m currently writing a batch of tunes for an as-yet-to-be-publicized project. I’m writing the melodies by making backing tracks, and then recording myself rapping the lyrics over them. After recording, I quantize the rhythms, and then run them through the Ableton vocoder. I set all the pitches of the carrier to the tonic, and listen back to the phrase on a loop while moving the pitches around until I find ones that I like. It works!