synthesizer

The synthesizer package provides functions for synthesizing melodies, chords, and stochastic compositions directly in R. Note that this project is a work in progress and will likely be missing features for quite some time and errors will propagate in e.g. counterpoint functions that will simply break the program rather than handled in the global composer-function call.

Table of contents

1. Installation

2. Synthesizer basics

3. Stochastic fugue generation

Installation

You can install the GitHub version with

devtools::install_github("shill1729/synthesizer")

You will also need to installl “seewave”, “tuneR”, and markovChains.

The former two can be installed via CRAN while the latter through GitHub

install.packages("seewave")
install.packages("tuneR")
devtools::install_github("shill1729/markovChains")

Setting up

How to set up for usage:

library(synthesizer)
library(seewave)
library(tuneR)
# Set player for Windows OS
setWavPlayer(shQuote("C:/Program Files/Windows Media Player/wmplayer.exe"))
# Sample rate for Windows
f <- 48000
# Set player for MAC OSX
setWavPlayer('/usr/bin/afplay')
# Sample rate for Mac
f <- 44100

Synthesizer basics

Playing a pure tone

How to synthesize and listen to a pure tone

library(synthesizer)
library(tuneR)
library(seewave)
library(markovChains)
# Set player for Windows OS
setWavPlayer(shQuote("C:/Program Files/Windows Media Player/wmplayer.exe"))
# Sample rate for Windows
f <- 48000
# Play middle C for one measure
x <- tone(midi = 60, rhythm = 1)
listen(x, f = f)

Synthesizing a melody

How to synthesize and listen to a melody with constant harmonics

library(synthesizer)
library(tuneR)
library(seewave)
library(markovChains)
# Set player for Windows OS
setWavPlayer(shQuote("C:/Program Files/Windows Media Player/wmplayer.exe"))
# Sample rate for Windows
f <- 48000
# Play a simple melody: C D E F E D C at middle C
midi <- c(0, 2, 4, 5, 4, 2, 0)+60
# In quarter notes
r <- 1/4
x <- tones(midi = midi, rhythms = r)
# Listen
listen(x, f = f)

Synthesizing a chord

How to synthesize and listen to a chord with constant harmonics

library(synthesizer)
library(tuneR)
library(seewave)
library(markovChains)
# Set player for Windows OS
setWavPlayer(shQuote("C:/Program Files/Windows Media Player/wmplayer.exe"))
# Sample rate for Windows
f <- 48000
# Play a Cm11 cluster
midi <- c(0, 7, 10, 0+12, 2 +12, 3 +12, 5 +12, 10+12, 12+12, 14+12, 15+12)+60
# For one measure
r <- 1
x <- tones(midi = midi, rhythms = r, type = "chord")
# Listen
listen(x, f = f)

Poisson processes

Subjecting the number of chord changes over time to follow a Poisson process, identifying the waiting times between arrivals of the Poisson process with the durations of the chords.

library(synthesizer)
library(markovChains)
library(tuneR)
library(seewave)
# Set player for Windows OS
setWavPlayer(shQuote("C:/Program Files/Windows Media Player/wmplayer.exe"))
# Sample rate for Windows
sampleRate <- 48000
# Possible chords to transition between: C, F/C, Dm7/C
chord_states <- list(c(60, 64, 67), c(60, 65, 69), c(60, 62, 65))
# Probability transition matrix
p <- 0.6
P <- rbind(c(0, p, 1-p),
           c(p, 0, 1-p),
           c(1-p, p, 0))
# Initial distribution
initial_chord <- c(1, rep(0, 2))
# duration of progression and mean number of changes
tt <- 10 # in measures
lambda <- 4 # changes per measure
# Default n = 4 for 4/4 time and bpm = 120
cprog <- poisson_chords(tt, lambda, chord_states, P, initial_chord)
listen(cprog, f = sampleRate)

Stochastic fugue generation

Two voice fugue with first species counterpoint

May stop on a counterpoint error, rerun it until it succeeds (won’t take long). Eventually, these errors will be caught and handled…
Also, this is mostly set up for subjects in a major key. The function will likely error out for minor keys, or just sound awful/unintended.

library(synthesizer)
library(tuneR)
library(seewave)
library(markovChains)
# Set player for Windows OS
setWavPlayer(shQuote("C:/Program Files/Windows Media Player/wmplayer.exe"))
# Sample rate for Windows
f <- 48000
# Number of statements
N <- 10
# Fugue subject
subject <- c(0, 2, 4, 5, 4, 2, 0)
# Rhythm of fugue subject
rhythms <- 1/8
# Try also random rhythms
# rhythms <- rexp(n = length(subject), rate = 4)
# Register of voices
registers <- c(48, 60)
# Probability transition matrices
P.voice <- rbind(c(0, 1),
                 c(1, 0))
I <- 0.1
R <- 0.1
RI <- 0.1
id <- 1-(RI+I+R)
P.transform <- rbind(c(id, I, R, RI),
                     c(id, I, R, RI),
                     c(id, I, R, RI),
                     c(id, I, R, RI))
keys <- c(0, 2, 4, 9, NA)
P.key <- rbind(c(0, 0, 0, 0, 1),
               c(0, 0, 0, 0, 1),
               c(0, 0, 0, 0, 1),
               c(0, 0, 0, 0, 1),
               c(0.7, 0.1, 0.1, 0.1, 0))
answers <- c(0, 2, 4, 5, 7)
P.answer <- rbind(c(0, 0.1, 0.1, 0.1, 0.7),
                  c(1, 0, 0, 0, 0),
                  c(1, 0, 0, 0, 0),
                  c(1, 0, 0, 0, 0),
                  c(1, 0, 0, 0, 0)
                  )
transition_matrices <- list(P.voice = P.voice, P.transform = P.transform, P.key = P.key, P.answer = P.answer)
# Generate stochastic fugue
s <- stochasticFugue(subject = subject, rhythms = rhythms, N = N, registers = registers, transition_matrices = transition_matrices, subjectKeys = keys, answerKeys = answers)
listen(wave = s$wave, f = f)