Sonification Project

My piece was created by sonifying the annual changes in the period of the Earth's rotation, measured in microseconds. There are three separate instantiations of that time series. Two (slaved to the left and right channel) are bowed instruments where the primary frequency is determined by summing a fixed component with another component controlled by the scaled time series, with random weights on each components. In addition, these components had modification to the bow location and pressure, as well as output gain. The third instantiation of the time series is a panning component comprised of a banded waveguide set to the tuned bar preset, with frequency intervals determined by the scaled time series. Because of the mechanics involved, none of the components are particularly reliant on the typical equal temperament scale.

Here is the recording:

If that didn't show up in your browser, the file is also available for download here:

Audio recording

Here are the ChucK files I used, as well as the data file itself. DataReader.ck is identical to the one provided online, and all the audio is run through Duo.ck (with its internal sporking behavior modified to fit my instrument choices and timing).

The data reader. The actual sonficiation code.

Data file (.txt format)

Answers to Questions

  1. The gain is squared in order to delinearize the changes in amplitude to better match human audio perception. As with light perception, we are better able to distinguish between the intensity of soft sounds than loud sounds with the same raw gain difference. Squaring the values for gain compresses the gaps for soft sounds and expands them for loud sounds, although it isn't precisely the same shape as the exponential function that would best match the decibel scale.
  2. Using MIDI keynums allows us to transform a linear time series into the exponential arrangement of pitch classes (at least, using the equal temperament system) in a computationally efficient way.
  3. If the update rate is increased or decreased significantly (as I did in my program, to be honest), it makes the resulting sound more clearly composed of discrete steps than a continuous sweep, increases the length of the program, and diminishes the smoothness obtained by using an envelope.