Music at all times, few were indifferent. In this case, someone, hearing a beautiful melody, just tapping to the beat with your fingers, some find it difficult to resist not to dance.
Scientists from the University Jyuvyaskyulya in Finland accidentally discovered a surprising phenomenon. It turns out that our dance style is unique, and the precision with which the computer determines the identity of the dancer in his movements remarkably high.
to find Out that the scientists were helped by the motion capture technology (motion capture technology). The same one that is used to create animation and special effects in movies.
“Actually, we intended to study something completely different. Namely, to check whether it is possible to move the manner to determine the music what genre of dancing people. This was done using a machine learning algorithm,” says the first author of the research paper Emily Carlson (Emily Carlson) in a press release from the University.
Finnish researchers involved in the experiment, 73 men and women. Volunteers were asked to just listen to music and dance as they want. To do this, participants were asked eight different styles (Blues, country, electronic music, jazz, metal, pop, rap and reggae).
According to the characteristics of the movements, the investigators tried to determine what kind of music listened to dancing. But unfortunately, experts created a computer algorithm determined the musical direction only in 30% of cases.
But scientists expected something more interesting. They were surprised to find that in 94% of cases the computer accurately recognizes the identity of the dancing. For comparison, if the algorithm had no data on the movements (that is, if he just “guessed” the identity of the participant), the expected accuracy would be less than 2%.
This observation prompted the authors to the idea that the manner of dancing is a kind of fingerprint, or “signature”. And it not depends on what tune dancing people, experts say.
However, some musical direction is still “confused”artificial intelligence. For example, dance under the metal computer will recognize the identity of the parties is less accurate.
Experts attribute this deviation with stable stereotypes about how the “need” to dance to that particular music. For example, most of the volunteers intensively shaking her head … which made them “faceless” for a machine learning algorithm, the scientists explain.
does the accidental discovery that in a short time to the software for facial recognition will be added, one specializing in the identification of the dance?
“We are much more interesting to realize that our observations are able to tell about the musicality of the person. Applications, for example, tracking people is not so important. Most of the studies raises more questions than answers, and ours is no exception. Whether “signature dance” unchanged for life? Are there cultural differences in the manner of dancing? Who more accurately determine the identity movements: artificial intelligence or humans?” says Emily Carlson.
All of these questions, the researchers plan to examine in the near future, which will improve the understanding of human musicality as a phenomenon.
a Detailed description of the experiment Finnish experts and the results published in the Journal of New Music Research.
By the way, recently the authors of the “News.Science” (nauka.vesti.ru) told about the alternative method to establish identity on the brain response to words. We also wrote about the unusual study proved the universality of music of different cultures.
in Order not to miss the most interesting news from the world of science and not only, subscribe to our groups in social networks: VK, Facebook, Twitter, Odnoklassniki. We have in Yandex.Zen.