AI reveals the complexity of a simple bird song

Photo of author


To the human ear, the songs of all male zebra finches sound more or less the same. But faced with the chorus of this simple song, female finches can choose the performer who sings most beautifully.

Zebra finches are found in Australia and typically mate monogamously for life, making this a high-risk decision for females. The zebra finch is among about a third of songbirds that learn a single song from their parents at a young age and sing it over and over again, raising the question of how females distinguish them from each other in choosing mates.

Listen to the song of a male zebra finch:

Scientists believe that most male songbirds evolved to sing a variety of songs to demonstrate their fitness. According to that theory, fitter songbirds will have more time and energy to work on their vocal styles and attract females with their varied vocal repertoire.

New research using machine learning shows that finches can follow a tune, but the way they sing it makes a big difference. Published on Wednesday in the journal reveals the complexity of a single zebra finch song and what songbirds might be hearing in the seemingly “simple” songs of their potential mates.

When researchers analyze bird songs, they often don’t listen to them, but instead look at spectrograms, which are visualizations of audio files.

“So I thought, ‘Hey, what humans do is look at images of these audio files. Can we use machine learning and deep learning to do this?’” said Danyal Alam, lead author of the new study and a postdoctoral researcher at the University of California, San Francisco.

Alam, along with Todd Roberts, an associate professor at UT Southwestern Medical Center, and another colleague, used machine learning to analyze hundreds of thousands of zebra finch songs to find out how they differed from each other and which variations were most attractive to female zebra finches. the songbirds. .

The researchers found one metric that seemed to catch the women’s attention: the distribution of syllables in the song. Women seemed to prefer longer “paths” between syllables. This isn’t something humans can easily pick up by listening to the songs or looking at the spectrograms, but based on how these algorithms mapped the syllables, the researchers were able to see them in a new way.

To test their hypothesis, the researchers took the findings to birds.

They generated synthetic bird songs to see if females preferred those with a longer path, and they did, suggesting that the birds’ target audience followed the same pattern as the researchers’ computers.

Listen to see if you can tell the difference between a synthetic finch song that doesn’t extend its syllables:

Alam and his colleagues also found that baby birds had a harder time learning long-distance song patterns than shorter ones, suggesting that fitter birds would be better able to learn them, the researchers said.

The hidden secrets of a simple bird song.

The study’s finding is consistent with what has been shown in other species: the more complex or difficult a song is, the more attractive it will be to females.

“Many signals in animal communication are intended to be an honest signal of some underlying quality,” said Kate Snyder, a Vanderbilt researcher who was not involved in the new paper.

For example, he said, if you look at a peacock, you’ll see that male birds with longer, more beautiful tails are better at attracting mates. Maintaining a tail like that is expensive for the bird, so it must be good at finding food and surviving in its environment to have time to dedicate to keeping its tail pretty.

“Learning takes a lot of time, energy and brain space,” Snyder said. Only the fittest males will have the time and energy to dedicate themselves to learning to sing.

Among finches, such work has been harder to detect…until now.

“We used to think of this one-song repertoire as maybe a simple behavior,” Roberts said. “But what we see is that it may be much more complicated than we previously thought.”

Leave a comment