Introducing {devoutaudio} - an experimental package for rendering graphics as audio


devoutaudio is an audio output “graphics” device - instead of rendering pixels, each graphics primitive is rendered as an audio snippet.

This is a testing ground for some graphics-to-sound ideas and is very experimental.

devoutaudio is written in plain R, and uses the devout package to interface with the R internals

Point-to-sound mapping

  • x position is mapped to audio channel position - points on the left of the plot are rendered mainly in the left channel of audio
  • y position is mapped to frequency - the higher the point the higher the frequency
  • size is mapped to duration - larger points have a longer sound


You can install devoutaudio from GitHub with:

# install.packages("devtools")


  • This is a very experimental device.
  • Only circles are currently mapped to a sound
  • Audio is only played on-the-fly and there is no facility to save yet.


In the following examples:

  • only audio is produced by the code show, but I’ve included the visual representation of the plots to assist in interpreting the audio
  • because the audio is never saved to file, the audio was captured on-the-fly from the MacOS sound output device

Example 1



Example 2

plot_df <- mtcars %>% arrange(mpg)

ggplot(plot_df) +
  geom_point(aes(mpg, wt, size = cyl)) +
  theme(legend.position = 'none')