I performed Elegy No.2, written in 2018 for violin and computer, with melodica at the SPLICE Inistute 2023. It is not a happy song, but I share what I can express only with music. Sarah Plum recorded the original version beautifully, but I have been playing the song as my solo shows since COVID.
If you own a melodica and want to play this, the score and SuperCollider file are available HERE. You don’t need to know how to use SuperCollider. The instruction to run the code is here. Please use the score as a guideline, and feel free to improvise.
The production of 847 Twins, the title track in the album Fan Art, is documented in four sections. The first section, Program, is a one-paragraph description of the music written for a concert booklet or album promotion. I share information and thoughts that may help listeners enjoy the music. The second section, Form, is for the creators who want to learn how I used electronic sounds in composition. The third section, Code, is for the technologists who want to learn how I designed the piece in SuperCollider, a code-based audio app. Links to the code are available here. The last part, Anecdote, has extra narrative relevant to 847 Twins but is optional to enjoy the piece.
847 Twins is a two-movement piece based on harmonic progressions of Prelude & Fugue in C Minor by J.S. Bach. An electronic remake of Bach is a well-known practice pioneered by Wendy Carlos and Pierre Schaeffer (Switched-On Bach & Bilude). I learned so much from reading and listening to their works. J.S. Bach is also my hero composer. Therefore, it seemed appropriate to dedicate a song to my musical cornerstones in an album about fandom.
Listen to the tracks linked below before reading the next sections.
Pluck and Blip, the two movements of 847 Twins, algorithms written in SuperCollider use the harmonic progression of the Prelude in BWV 847. The downloadable code, 847_Pluck.scd, generates randomized voicing patterns played by a guitar-like synth. Below is a step-by-step explanation of how the composition process.
Design an electronic string instrument. Each note of this instrument is detuned at a different ratio every time the string is “plucked.” The note’s duration, dynamic, string stiffness, and pan position also vary randomly.
Using the instrument in Step 1, strum a chord with notes at a measure in BWV 847. Unlike a guitar, a strum of a chord can have multiple pan, accents, and note durations due to the randomization in Step 1.
Each measure of BWV 847 is played four times before advancing to the next measure.
Add a bass part with gradually increasing loudness. It plays the lowest note in the corresponding measure.
Add the intro and the outro for a better form. They are not quoted from BWV 847.
In short, the first movement of 847 Twins is a reinterpretation of BWV 847 featuring an imaginary string instrument and a synth bass. I loved how Bach created exciting music with a predictable rhythmic pattern. The key was harmony and voicings. I wanted to emphasize that aspect with an additional layer of dynamics articulations in Pluck. The added bass line, which imitates the “left hand” of basso continuo, fills in the low-frequency spectrum of the piece. The bass part is best experienced with a headphone or a subwoofer.
Mvt II. Blip
The first movement lacked elements of counterpoint, so I tried to make an electronic polyphony in the second movement. In Blip, each measure has 3-6 parts playing different phrases derived from a measure in BWV 847. The phrase shape, the number of voices, and articulation are determined randomly at every measure and create a disjunct yet relative form. Schaeffer’s Bilude explores this idea by combining piano performance and recorded sounds.
Below is my process of creating a random phrase generator. Please run 847_Blip.scd to hear the piece.
Create a list of pitch sets by reducing repeating notes in each measure of BVW 847.
Make three different synth sounds.
Make a phrase generator that uses the list in Step 1 and synths from Step 2. The instrument choices, phrase length, note subdivisions, and articulations are randomized. The SuperCollider code also has the option to generate a rhythmic variation (i.e., insert rest instead of a note).
Make a polyphony generator that spawns the phrase generator described in Step 3. The number of polyphonic voices and their octave transpositions are random.
Play and record Step 4 twice. Then, import the tracks to a DAW. Insert a reverb plugin on one track. The reverb should be 100% wet.
The algorithm described above creates different timbres, polyphonic patterns, and the number of voicings at every measure. Furthermore, every rendition of the SuperCollider code makes a unique version of Blip. One measure can be a duet of two-note phrases, and the following measure can be an octet of eight phrases played in a four-octave range. The room sound created by the DAW reverb plugin doesn’t reflect the source, but it sounds similar enough to be heard as part of a whole.
Code
Mvt I. Pluck
The SuperCollider file for Pluck consists of seven parts. Please download and use 847_Pluck_Analysis.scd to hear and modify each part. Make sure to run the line s.options.memSize=8192*16 to allocate enough memory.
SynthDefs: SynthDef(“Gtr”) uses a Karplus-Strong physical model with controllable pan, frequency, stiffness, amplitude, and duration. SynthDef(“Bass”) makes a sinusoid tone with a percussive amplitude envelope. The UGen Lag.kr smoothens the sharp transient of the amplitude envelope.
~onenote: this function uses two SynthDef(“Gtr”) to create a detuned note. The amount of detuning is randomized along with other parameters of the SynthDef.
~stroke: this function creates instances of ~onenote with pitches specified in the ~chords array. ~chords is a collection of all the notes in the Bach Prelude, categorized and indexed by measure number. The order of the notes in a measure is random. ~stroke plays the chord in sequence or reverse to simulate a guitar’s up and down stroke motions.
~strums: this function continuously triggers ~stroke. The global variable ~pulse determines the tempo. ~strumsend function is used once for the ending.
~clock: this function changes the chord progression at time intervals set by the global variable ~mdur. It also changes the parameters of ~strums by altering the values of global variables ~mm, ~accent, ~volume, ~notedur, and ~stiff. Note that both ~strums and ~clock functions must run simultaneously for a correct chord progression.
~bassline: this function plays SynthDef(“Bass”) a few seconds after the start of the piece. It uses the if condition to change the rhythmic pattern. The line pitch=~chords.at(count).sort.at(0) picks the lowest note of each measure as a bass note.
SystemClock: this scheduler syncs ~strums, ~clock, and ~bassline to play a version of Pluck. Every rendition of SystemClock will make a new variation of the track.
Mvt II. Blip
The SuperCollider file for Blip consists of four interconnected parts. Please download and run 847_Blip_Analysis.scd to hear each part.
SynthDefs: The three SynthDefs, PBeep, TBeep, and SBeep, are all slightly detuned percussive instruments featuring a classic oscillator waveform, such as sine, triangle, and pulsewidth.
~phrase: this function creates a short melodic pattern based on pitch sets received from global variable ~arp. It controls which SynthDef to use, amplitude, phrase length, note duration, and transposition. The last two arguments activate or deactivate that random rhythm generation and arpeggio pattern variation.
~section: this function duplicates ~phrase. The number of ~phrase and octave transpositions are randomized. The function also makes further variations on amplitude, note duration, and panning.
The Routine in the last section uses the ~piece array as a cue list with details on when and how to trigger the ~section. The array ~chords is a list of all the notes in corresponding measures of the Bach Prelude. The Routine also sends a changing pitch set from ~chords to ~phrase via the global variable ~arp.
Anecdote
847 Twins does not use the Adagio section of the Prelude and Fugue. When composing the first movement, I could not transition from a constant 16th-note drive to a free and improvisational ending. I tried to address this incompleteness by writing a complementary movement, Blip, but it did not work out. I made a satisfying solution six months after completing 847 Twins by incorporating an instrument I could improvise aptly and freely. Nim6tet, the sixth track in Fan Art, has six layers of no-input mixer improvisation guided by the chord progressions of the Adagio section. It shamelessly shows off no-input mixer sounds I can not create with other instruments.
It took many attempts in the period of 1.5 years to finish three tracks about the first half of BWV847. The electronic interpretation of the Fugue part is a puzzle yet to be solved.
End Credits, a fixed media composition included in the Fan Art album, is documented in five sections. The first section, Program, is the program note to be included in a concert booklet or album promotion. I share information and thoughts that may help listeners to appreciate the music better. The second section, Form, is for the creators who want to learn how I used electronic instruments to create a complete piece. The third section, Code, is for the technologists who wish to understand how I coded the piece. Musicians familiar with code-based apps like SuperCollider and Max will benefit from analyzing the code. In the fourth section, Inspirations, I share why I chose to write the piece. The content is too personal to be on the program. The last part, Uniquely Electronic, is a bonus section featuring sounds and ideas I could not express in an album format. Fan Art is available as streaming stereo tracks, but they are originally designed for multi-channel sound installation. The last section provides resources to realize songs in Fan Art at full capacity.
End Credits is an algorithmic composition based on the harmonic progression of Debussy’s Clair de Lune. The SuperCollider code written for the piece generates notes with unique overtones, and the overall sound reminds me of organ music at viewings. My friend and I joke about writing each other’s farewell music, and I got one for him now. If he doesn’t like it, I will use it as my exit theme.
End Credits uses the harmony of Debussy’s Clair de Lune. The downloadable SuperCollider code, EndCredits.scd, makes sound according to the following instruction.
Choose a list containing all notes present in measure x.
Scramble the order of the notes.
Play notes at random timing. There is a 50% chance of two notes being played simultaneously.
When all notes in the list are used, move to measure x+1.
Repeat steps 1-4 in slow tempo (quarter note = 3.6 seconds). End Credit uses harmonic progressions from mm1 to mm27 of Clair de Lune.
EndCredits.scd code also generates each note according to the following instruction.
Make a sine tone with randomized slow vibrato and tremolo using two LFOs. Randomize amplitude envelope, LFO rate, LFO amount, and pan positions.
Make a single note by combining 5 sounds made in Step 1. Then, randomize each note’s positions, frequencies, and pan positions to make a slightly detuned note with a wide stereo image.
Make a single note with a random number of overtones using the note generated in Step 2. The note’s duration is also random but is almost always longer than a quarter note (3.6 seconds).
The resulting sound is an imaginary organ capable of changing the stops at every note. The instrument also seems to have multiple sustain pedals.
The compositional objective of making End Credits is akin to minimalism. I wanted to create a simple process that yields unexpectedly delightful sounds. So I simply made an ambient piece using additive synthesis and traditional harmony with computer-aided instructions. There’s no new technology or concept, but we create new sounds by combining old ideas.
Code
EndCredits.scd, the SuperCollider file I made to generate the album version of End Credits, has the following sections.
SynthDef (“Cell”): makes a sine tone with controlled random values
~note: make a note event using SynthDef(“Cell”)
~chords: an arrayed collection of note numbers representing harmonic contents.
~event: play one measure using ~note with pitches in ~chords
SystemClock: play the music
To make sense of this section, open EndCredits_Analysis.scdon SuperCollider and refer to the code while reading the next sections. The analysis .scd file has simplified working codes.
SynthDef(“Cell”)
End Credit uses one SynthDef featuring two LFOs, one ASR envelope, and one stereo sine tone generator. By playing two instances of SynthDef(“Cell”) with slight pitch differences, we can make a simple detuned sound. The codes in section //1. “Cell” without randomness shows the simplest form of the instrument. The actual SynthDef used in the piece is in //2. “Cell” with randomness. It applies ranged random values to give varieties in amplitude envelope’s attack and release time, LFO’s frequency, phase, amplitude, and pan position.
~note
~note is a function with the following parameters.
~note.(pitch in MIDI number, duration (sec), volume (0-1), number of overtones);
By providing a number for each parameter, ~note creates a tone with a varying number of detuned overtones. As we can observe in //3. ~notes, loops (.do) and ranged random number generators (rrand) were extensively used. Try to run the following line on SuperCollider to hear the difference. Notice that the sounds are not identical when the codes are re-evaluated.
~note.(50,10,0.3,1); //no partials
~note.(50,10,0.3,5); //some partials
~note.(50,10,0.3,10); //many partials
~chords and ~event
~chords is an array of interval values representing the notes in a measure in Clair de Lune. As evident in //4. ~chords, method .scramble is at the end of every measure to randomize the sequence order. For analysis purposes, only three lists are inside ~chords.
~event plays ~notes according to the pitch choices in ~chords. One measure in End Credit is thus generated with the following parameter
~event.(measure index number, note duration factor, amplitude, overtones, harmony probability (0-1.0))
As we can observe in //5. ~event, the note duration factor is a multiplier for each note’s duration. The larger the number, the longer the note duration, resulting in a sustain pedal-like effect. The overtone amount also gets a slight randomization for variety. The last parameter, harmony probability, can control the chance of the following note being played simultaneously.
SystemClock
Codes in //6. SystemClock is responsible for putting everything together to produce audible sounds. The section uses Routine to make ~event go through all the measures provided in ~chords. SystemClock provides a 2-second silence, a little pause before listening to everything.
Inspirations
I composed End Credits n February 2021, the end of strict COVID-19 isolation days. I must have been listening to Debussy recordings often to keep myself together. Clair de Lune’s lowest note, Eb in mm15, felt like the most beautiful piano note then. The timing was perfect, and it resonated with the piano’s body and the listener’s mind. I wanted to recreate that in an electronic music context. So, I created a sound that imitates the slight detune of the low range of the piano. Then, perhaps due to COVID blues, I instructed the computer to play those sounds in blurry and slow motion.
Uniquely Electronic
End Credit has two playback modes: ~onetime or ~infinite. The one-time version with a fixed duration (8:26) is available on major distribution platforms for listening, and any media player can play it. The installation version goes on indefinitely with varied timbre, timing, and duration. The listener will need to runEndCredits_infinite.scdon SuperCollider. Let it run for hours on solemn and not-so-happy occasions!
Self-assessment of solo set performances from 2011 to 2022
I analyzed instruments, gear, and repertoire of my solo electroacoustic sets from 2011 to 2022. For the sake of discussion, I categorize a set performance as a long-form performance (about 20 to 50 minutes) by artists without interruption or intermission. I find preparing, performing, and refining solo sets to be one of the best practices for live electronic music skills. In this article, I analyze the changes in hardware and repertoire in my solo electroacoustic performance by comparing video recordings of past shows. I hope the analysis serves the readers and me on what could be worth exploring in live electronic music.
Here are three video recordings of my solo sets as a point of reference.
The videos represent my performance practices in three periods in three cities I worked and lived: Philadelphia (2009-2014), Oberlin (2014-2016), and Detroit (2016-current). The oldest solo set video I could find is from 2011 in Seoul, but the above videos are unabridged and have direct outs from the house mixer.
Performance Preparation
Practicality and flexibility matter in a set performance. Venues have sound systems with varying designs, and sharing the stage with other artists doing a set is common. Therefore, I choose pieces according to the external limitations I cannot control. My gear is compact and travel-friendly to set up and strike quickly, regardless of the venue’s PA capacity. I plan 15 minutes or less to set up, get ready to go on stage, and carry an extra direct box and cables.
The most efficient setup could be solo laptop performance. But I am not inclined to present in that format as the audience cannot see movements behind the computer screen. The gear placement reinforces the visual cues in my shows. I value establishing a connection between what I do on stage and what the audience hears. In the reference videos above, the laptop is on the right side of the table. Most sound-generating objects and body actions are in the center and have an unobstructed view. When possible, I place the front panel of electronic instruments observable to the audience. For example, the picture below is my setup in October 2022. I put the gear on a piano bench for the audience to see the fingers moving across the buttons and sliders. The laptop created sound, but it was tucked below the bench. I launched the SuperCollider patch before the show and did not need to touch or look at it when performing. Not shown well in the picture is a small three-channel mixer on the floor. Like the laptop, there was no need to touch the mixer after the sound check, and it did not need to be on the table along with the instruments.
The 2013 Set
The diagram below depicts the connection of the hardware used in the set. The three boxes in the diagram contain the no-input mixer patching method, a list of SuperCollider patches, and compositions incorporated in the set. SuperCollider is the sole software I have used in solo performances since the early 2000s. No-input mixer patching details differ from artist to artist, and it is worth documenting my preferred patching for comparison. I will use the diagram in the same format for the 2016 and 2020 sets.
Hardware choices depend on the repertoire. The title track of the performances in 2013 was Toccata, an improvisational piece featuring a wooden board, various acoustic objects, and a live processing SuperCollider patch. A combination of a contact microphone and a small diaphragm condenser mic captured sent audio signals in the air and the board. The condenser mic doubled as an audio input for other pieces like Retrace, Introvert, and Elegy. Retrace is my first SuperCollider composition incorporating an acoustic instrument. The solo set format allowed me the repeated performance of Retrace and gave me multiple chances to refine it. The picture below, taken at a 2013 Indeterminacies series in Tennessee, is an example of a typical performance layout.
The microphones were connected to an audio interface with four inputs and outputs. Audio inputs 3 and 4 allowed me to connect outputs from a no-input mixer. In 2013, the sounds of the no-input mixer and the synths took up a small portion of the show. I used them for the first time to create videos for the 100 Strange Sounds project and was not proficient in performing them. From 2013-14, many objects I experimented with in 100 Strange Sounds became part of Toccata. It refined the piece composed in 2009 for four or more years.
Video projection was also part of a performance in 2013. Both Introvert and Elegy have accompanying videos, and I often used the laptop’s built-in camera to project the hand movement on the wooden board. The 2015 performance at New Music Gathering is an example of such a set. The on-stage live video reinforces the connection between what I do and what the audience hears, but not all venues can accommodate large-screen projection systems. Setting up a proper video meant extra tech time, a possible nuisance for tech people and other artists on the same show. The reduction in practicality led me to retire the video features from the set gradually.
The 2016 Set
I avoid unintentional silences between the pieces in a set performance. In 2013, I often played pre-recorded sounds while adjusting SuperCollider patches for the next piece. In 2016, all transitions became superpositions of the end of a piece and the beginning of the next one. The crossfade time got longer and smoother with more experience, blending multiple works. This approach stimulated new compositions combining two or more previously featured instruments. Consequently, my set became a single long-form improvisation featuring all the instruments I could carry comfortably. The 2016 WOBC video recording serves as an example.
Out-of-town gigs outnumbered in-town opportunities when I lived in Oberlin. Traveling with a carry-on bag full of gear became burdensome, and thus I sought to develop a set with as few instruments as possible without degrading the quality. In 2015, I had an opportunity to perform on a double-decker tour bus. The setup time was short, and the performance space was small due to the particular nature of the gig. So I chose to abandon the laptop, the instrument I am most skilled at, and performed with a no-input mixer and a monophonic synthesizer. The success and fun I had in the gig encouraged me to add more non-computer elements to the set.
The 2020 Set
By 2016, I felt at home performing with a contact mic and found objects. But doing so felt less exciting and challenging. In contrast, my interest in drum machines and MIDI controllers grew. The resulting pieces were Cobalt Vase and Page Turner’s Agony. By combining Cobalt Vase with no-input mixing, I composed Func Step Mode. These three pieces are currently the main ingredients of my solo set of electronics-only improvisation. The video made for La Escucha Como Acción’s COVID online performance series is an example.
The current set does not include visually expressive works like Toccata. Microphones and found objects are absent, limiting sonic and visual possibilities. But I gained mobility and a chance to showcase my skills on specific instruments in return. The set sends mono output and can work without an audio interface. The output choice is efficient but could be risky in a genre that values high-fidelity and multi-channel audio. But I identify the most with the sound of the current set.
Findings from Analysis
Reviewing a decade of set performances was an opportunity to evaluate what I value most in live electroacoustic music. I value practicality and refinement. I accept that practicality gets priority over aesthetics in my music. Some pieces are no longer in the setlist because they involve more physical labor and are prone to technical errors. The longest surviving instrument over a decade, besides SuperCollider, is a no-input mixer: it takes a short time to set up and is immune to software updates. This reliability led to more time spent with the mixer. A deliberate decision on one instrument is worth noting in technology-based performance, in which one can access an uncountable number of synthesizers and controllers. Currently, I feel proficient at performing a no-input mixer. I am developing a similar feeling toward Korg Volca Beats. Efficiency affects aesthetic choices in my music.
I recommend designing, executing, and revising a solo set for electronic musicians. Preparing and practicing sets builds muscle memory and opportunities to overcome weaknesses. Compositions featured in a set get continuously refined with repeated performances. Combining and remixing works in a set often inspires new compositions. The refining process is a luxury for compositions written for others, but set performances demand it by nature.
