Visualizing Birdsong
Avian calls like you’ve never seen them before.
As you study this image, you are, in a sense, seeing sound. Those agitated turquoise lines and violet ellipses represent a four-second recording of the song of the superb fairywren, a common Australian songbird. The image is just one of a colorful collection, created by software programmer Mark Fischer, that depicts songs produced by cetaceans and birds.
See this article's accompanying photo gallery
Birders in particular might be familiar with a type of sound visualization known as a spectrogram. To generate a spectrogram of birdsong, a scientist runs an audio file through a computer, which applies a mathematical algorithm to create a visual representation of the tune.Fischer uses a similar approach, but with a different mathematical mode of visualization, called a wavelet transform, employing a computer program he wrote himself. The result is more descriptive, he says—while the spectrogram does well at illustrating sustained notes, the wavelet transform excels at capturing short bursts of sound, like clicks or chirps.
While this fairywren image cascades like an embellished musical score, other melodies in Fischer’s gallery produce images that are downright psychedelic in color and pattern. Bold neon yellows or electric blues clash and meld into zany, quivering shapes. Fischer lets both math and his artistic judgment guide his choices. Here he opted for a color scheme to complement the singer’s plumage.
Fischer thinks his method could be useful to bioacoustics—which studies how organisms produce and receive sound—by providing insight into such avian conundrums as the “cocktail party” problem, when multiple birds croon at once. Studying a wavelet transform of the chatter might suggest how one bird communicates with another above the noise, Fischer says. Regardless of application, his work sings.
This story originally ran in the March-April 2013 issue as “See a Happy Song.”