AI has come a long way in the past decade, and now it’s about to take on one of nature’s most impressive feats: bird song. A team of researchers from the University of Washington has developed a new AI system that can recognize and generate bird song with unprecedented accuracy. The team’s findings, published in Nature Communications, could open up new avenues of study into the complex vocalizations of birds, as well as provide insights into the evolution of language. The AI system, called Chorus, is able to recognize and generate bird song with 97% accuracy, which is much higher than previous attempts. Chorus is based on a deep learning technique called “convolutional neural networks,” which allow the AI to learn from a large set of bird songs. With this knowledge, Chorus is able to accurately identify both the species and the individual singing a given song. In addition, the AI can generate new bird songs that are indistinguishable from real ones. This could be a huge step forward for the study of bird behavior and the evolution of language.

Read Full Article Here

source: Phys.org