Throwing Darts
The previous post in this series discussed concepts of microtiming in groove based and improvised music. While models for the analysis of sub-tactus onset events and methods for examining such events at the phrase level exist, such approaches are in fact more forensic than generative as they do not engage with preparatory microtiming, the embodied process of initiating a note onset event. My goal is to create an organic drum machine with Max/MSP/Jitter using algorithms that elevate the human reality of musical timekeeping above the clock-based fantasy model of rhythm found in current drum machines. This task presents unique problems at every hierarchical level of rhythm from microtiming to form. Given that human timekeeping employs various brain functions, constantly refers to the surrounding musical environment, and relies upon complex brain/body/instrument relationships for expression, a truly functional organic drum machine would require significant amounts of data gathering using human performance models to build effective algorithms.
Such data could not be taken from existing recorded performances; these too would be forensic rather than generative, and the goal is to make the organic drum machine a human-like generator of rhythmic patterns. Ironically, it seems that some data should be created in controlled settings where musicians are recorded playing along with a mechanized rhythm of some kind (how else can the exact location of the tactus be known?). Such performances could be analyzed and the data used to implement appropriate formulas for the machine to use. But how should such analysis be conducted? What is the relevant data? Individual musicians respond to and enact musical timing differently; we all have personal ways of understanding, expressing, and adding to the rhythmic texture of music during performance. The performer’s perspective on and concept of his/her own process should be included in this analysis. In this way information pertinent to the enactment of rhythm can be reinforced and understood from the player’s point of view. Much of my current work explores the boundaries between individual and shared musical understanding, and in my composing, programming, and improvisation I seek to discover how my brain processes and remembers music towards the goal of creating a truer personal expression of music. Research leading to the development of an organic drum machine presents a rich forum for investigation of the simultaneously disparate and common narratives, strategies, and emotions we express through rhythm, and an opportunity to create computer-based music that is deeply rooted in human methodologies.
The following 10-second excerpt from my composition Descarga (3:50-4:00) presents several rhythmic events that further complicate concepts discussed in this series:
Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.
At this point in the piece the computer had just finished transitioning from one rhythmic cell, called “jr,” to another one called “2gin.” While the rhythms played during the transition (not excerpted) let me know we would arrive at 2gin, I had no way of knowing which exact 2gin mini-composition would start at the downbeat where this excerpt begins. For each rhythmic cell in Descarga there are two or more possible sections the computer randomly chooses to play, each with a different set of patterns and groupings. As it happens the computer and I ended up playing more or less in unison. 2gin is a seven-beat rhythmic cell based on 3:2 rumba clave. In this mini-composition the 2gin pattern is introduced by 14 repetitions of a dotted-eighth/sixteenth type pattern. In the excerpt above improvised bass and sampled percussion sound a set of five more-or-less unison pairs of dotted-eighth/sixteenths. During the next nine beats the percussion holds steady as the bass pattern wanders around, finally returning to almost-unison with the percussion’s last two notes.
Taking a more detailed look, the percussion pattern alternates between two sounds-low and high-and never deviates in its pattern or order of pitches. While my improvised pattern does deviate away from unison with the percussion, the bass pattern maintains a similar low-high (A-E) gesture. In the last half of the excerpt 5 low-high bass patterns occupy an indeterminate rhythmic space more or less unrelated to the computerized percussion pattern. The bass finally returns to match the percussion pattern, but now the bass pattern is reversed high-low in relation to the computerized percussion’s low-high. The overall effect, while certainly disjoint in the last half, ends in a way that implies I knew how to end my wandering two-pitch pattern and was intending to reverse the order of pitches in relation to the computerized percussion: The computer pattern and my improvisation both end on the downbeat and pause. However, I know that during the improvisation I was not able to keep track of how many computer percussion patterns were elapsing as I varied my rhythm. While this excerpt represents an improvisational and rhythmic success for me–a cool idea that worked out well–the mechanisms I used to pull it off were far from surefire.
Chart 1 shows the timing of all the onset events in the excerpt. Time progresses from left to right. The lower two symbols represent the computer percussion (beats and upbeats, on a grid), while upper symbols show the two-pitch improvised bass pattern. At first the bass’ pattern closely matches the computer both in rhythm and gesture: low pitch on the beat, high pitch on the upbeat. The bass beats are almost simultaneous with the percussion beats while the upbeats slightly anticipate the percussion upbeats, swinging the rhythm. This initial unison effect establishes two relationships between improvisation and computer rhythm: first that they are rhythmically similar, second that low and high onsets mark the same type of beat subdivision. At beat six the improvised line begins to depart from the computer’s pattern. While the bass and percussion patterns grow apart, the beat/upbeat low/high relationship allows the bass pattern to maintain a sense of gesture so that when the final pitches of the bass are low-high against the percussion’s high-low expectations established in the first half of the excerpt are defied.
Transcription of these onsets presented a challenge. The sense of gravity created by both the computer percussion and elastic bass pattern invite various divergent notational strategies. In addition low frequencies and room sound reflect and muddy the attacks. I finally chose to use a close-mic recording of the bass performance, slowed the excerpt to half tempo in Max/MSP, filtered the audio to emphasize the bass’ overtones at close to 880/1320 Hz, and used pro tools to locate transients near audible onsets. The animation shows how the pattern unfolds. Analysis reveals that although the bass’ pattern in the last half of the excerpt does present certain rhythmic relationships to the computer percussion what I am mainly doing is slowing my pitch A beat pattern (chart 2). In addition, with each ritardando low-high cycle I am shifting the location of E, the high pitch, so that it is more “centered” within the the low A inter-onset interval (IOIs) around it. As my pattern slows, its rhythm also becomes more evenly subdivided (chart 3). (all data from analysis here)
Careful analysis plus my perspective as the performer leaves little doubt that the pattern in the last half of the excerpt is gradually slowing in relation to the computer percussion. I am ignoring the accompaniment rather than playing off of it. However, as I successfully slowed down and shifted the upbeats in my improvisation, I’m sure I lost track of where I was within the longer metric system. No doubt I knew it was the seven beat 2gin pattern, and this percussion pattern would soon be expanded to include other groupings, but in the last five seconds (if not longer) of this excerpt I am lost and don’t know when beat 1 will return. The final pair of bass pitches that return to (almost) match the percussion and invert the low-high pattern are stumbled upon rather than intentionally played with the downbeat of the 2gin pattern. As I slowed down my As-Es I could hear percussion patterns passing by but was not counting them. The final two bass pitches in the excerpt, also the most closely spaced, show how I was brought out of my meandering reverie. The final low A follows the slowing down trend, but immediately after it I hear the computer percussion’s high pitch and, as quickly as possible, (so quick my onset is early) play my last high E so it matches with the reliable percussion pattern that, even without counting, I know the shape of. This could easily have happened three beats earlier when my A also aligned with the percussion or, after the percussion momentarily stopped (right after the end of the excerpt), I might have continued playing. In hindsight these possible outcomes seem less elegant than what I did play.
I was lucky to end up back with the percussion. In the seconds before that happened I was ” throwing darts,” playing kinda slowing down and strangely subdivided pitches that landed somewhere within the accompaniment, and the processing required to play those pitches made me forget where 1 was until something (that last high drum) snapped me back into lock step with the rhythm. If I’d stopped two beats earlier that might have sounded incomplete. If I’d continued playing when the percussion stopped who knows what would have happened. This excerpt, although not an example of a grooving pattern, does contain the kinds of complex “binary” (like high and low pitches in bell patterns) rhythmic data I enjoy working with as a composer, and does relate in some ways to the kinds of patterns I hope to create with the organic drum machine. While this analysis does not bring me much closer to a set of guidelines for analyzing human rhythmic performance, it does present some new concepts for me; the idea of shifting a note’s “centeredness” within other notes, faster or too fast reaction times at clutch moments, and the possibility that recorded rhythms which in hindsight sound well thought out may owe more to chance than extra-nimble musicianship.
Microtiming formulas will clearly be a significant part of the organic drum machine’s programming. It’s possible that phenomenon discovered through controlled study of performances patterns of microtiming can be indexed to performers’ own sense of their rhythmic expression, and, by extension, lead to distinctly human ways to program in the digital environment. Other timed natural processes, like plant growth, stride pace, respiratory rhythms , etc, show self similarity and cyclical patterns when studied at different time scales. Perhaps human timing in music contains self-similar patterns at various temporal levels as well. The next post in this series will make a clearer definition of what the organic drum machine should do and establish basic goals for the project.
